EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: electronwaster on February 02, 2012, 10:59:35 am
-
Hi there,
I bought an altronics branded 25-0-25 300VA toroidal transformer from the local tip shop (recycling centre) for $1. It appears unused as the tips of the leads are still enamelled - I'd like to build a bench power supply with it. I'm currently using (playing around with) a modified antec PSU with multiple outputs but no voltage adjust or current limit (except the built in thermal overload or whatever), and I would really like these features, so I think I may use the transformer as a basis for a nice linear bench power supply.
I have been watching the recent video series here on eevblog about power supply design, but there is way too much going on in those circuits for me to be able to understand everything that's going on. Also - I think after reading the pros and cons of linear vs smps, I would like to stick to linear design and 10-turn pots for my first supply, keep it simple principle and live with or work around the downsides like massive heat dissipation.
I think I've got it right that a 25-0-25 transformer will put out 25*sqrt(2) VDC per output after rectification and smoothing, so +/- 35V DC. At 300VA, this could supply around 4A, though that is a hell of a lot of power to dissipate with an LM317 or LM338 when the voltage dial is turned down to 0.1V and lots of current is drawn, around ~280W or so. I will have to really figure out how pass transistors work with adjustable regulators, and then make my own short circuit protection etc. I may even need to think about temperature-switched fan cooling, as I expect to be drawing about 50mA most of the time and would like it quiet, but would like to know that things are being cooled properly if I do need to draw on those extra amps.
I have seen many circuit diagrams all over the internet, and they all seem to differ slightly. Some don't have panel meters, though I could figure out how to add those to any circuit, some don't have +/- supplies, though I could just build two identical circuits, because I have 4 leads on my transformer secondary, some designs don't have current limiting which is a major thing that I want, but don't understand how it works. Another feature I think I will need is the ability to go down to 0v output, which again they don't all have, and again I don't quite understand. It's very hard to find one diagram that fits the transformer I have and the features I want! I will keep searching of course.
The main reason for posting was to show off about my toroidal transformer, the other reason was to ask if I'm heading in the right or wrong direction with planning to use this transformer. Is this kind of voltage and current output too high for a linear circuit, practically? I think the high power output will be useful to me if not now then later (building/testing audio amps, charging 12v or 24v battery banks etc.), but if it's going to become impossible for me, a beginner, to make, then maybe it's better to start smaller. What do you all think? Should I be looking to make a nice stereo amp with it instead?
electronwaster
PS: As usual - the main reason for building rather than buying is not to save money, but to gain experience and learn something! i.e. I realise it will probably cost me more money.
PPS: With an LM317 or similar, to get a variable output right down to 0V can I do this? -> If I can normally get 1.2-40V, for example, can't I just put two diodes in series from the ground, and the diode forward voltage drop will give +1.4v which I can use as my new output ground?, and so get -0.2 - 38.6V for example? Isn't this easier than a second negative power supply - or do I misunderstand something?
-
The only way to make use of the full 300VA is to make a switching power supply, but it is really hard to build a switching variable power supply that comes close to matching a linear regulator over the full voltage range.
So definitely stick to a dual linear power supply, and just lower the current to whatever you can comfortable dump into a heatsink.
If you leave the circuit aside, and just stick to getting rid of power, then with a $27 heatsink that is convection cooled and that is typically below 70 degC (the maximum you should have for any surface you can touch), then we are probably looking at a dual 30V supply at 2A, or with a voltage doubling rectifier, a dual 60V supply at 1 A.
Now even though the transformer was only $1, the rest of the parts including a nice strong steel case strong enough to hold the transformer will almost certainly come to over $100. Is that OK? If you do a good job, then an equivalent commercial linear supply could easily be over $1000. Also forget about how simple you can make a power supply with a LM317. By the time you have added everything else you need, along with all the wiring, it is a big project. I actually don't think you will want to touch a LM317. You will want a power supply which works perfectly down to zero volts and zero amps, and that has separate voltage sense wires that go to the output for best regulation. If you are smart, you can do all that with less design effort then a LM317-based supply.
If this sounds like what you are thinking about, I can suggest some next steps.
Richard.
-
Unloaded voltage will be slightly more than 35VDC, since the transformer will probably supply more than its rated voltage when unloaded. Large transformers are not as bad as the tiny ones in this regard, probably only a few volts for 300 VA (see data sheet). Not an issue for dissipation, as long as you don't exceed the max voltage ratings of your components.
If it has two separate secondaries, I would probably just make two isolated positive power supply sections that you can put in series if necessary. If it's a split secondary, then it's not as easy. Designing a negative regulator doesn't seem worth the effort, it takes extra design time and increases the BOM size.
I would arrange for a negative supply to the regulators/opamps. It doesn't seem worth to choose exotic parts like in Dave's design, you can just use a small transformer to supply the negative voltage or wrap some extra windings around the toroidal transformer. Note that if you use an external pass transistor with a three terminal regulator, you will lose the over current and overheating protection of the regulator, so make sure to take care of that.
Putting diodes in series will increase the output impedance and degrade load regulation, especially at low voltages. For example with a single 1N4007, at 1A the forward voltage will be about 0.9V typical, and at 10 mA it would be 0.7V. That means that 0.5 V at 10 mA will suddenly be 0.9 V at 1A. That's some pretty horrible load regulation.
-
Note that if you use an external pass transistor with a three terminal regulator, you will lose the over current and overheating protection of the regulator
isnt that possible with power resistors division to divide the current properly so that at some current fraction into the regulator, will trip the protection? say with pathetical example 0.9ohm in series with regulator and 0.1ohm at pass transistor, the regulator will trip say when 1A goes through it, while during the time 9A goes through the pass transistor, so cant we say we have 10A circuit protection now?
For example with a single 1N4007, at 1A the forward voltage will be about 0.9V typical, and at 10 mA it would be 0.7V. That means that 0.5 V at 10 mA will suddenly be 0.9 V at 1A. That's some pretty horrible load regulation
cant we use the sense/feedback wire as amspire said to avoid this?
voltage sense wires that go to the output for best regulation
-
Hi Richard,
The only way to make use of the full 300VA is to make a switching power supply, but it is really hard to build a switching variable power supply that comes close to matching a linear regulator over the full voltage range.
This is what I thought - I've looked at the datasheets of the "jellybean" MC34063 smps, and all the components are finely tuned for one particular output voltage, so you have to take the worse case for each component and every output voltage is then a bunch of compromises as far as ideal components, and efficiency.
So definitely stick to a dual linear power supply, and just lower the current to whatever you can comfortable dump into a heatsink.
If you leave the circuit aside, and just stick to getting rid of power, then with a $27 heatsink that is convection cooled and that is typically below 70 degC (the maximum you should have for any surface you can touch), then we are probably looking at a dual 30V supply at 2A, or with a voltage doubling rectifier, a dual 60V supply at 1 A.
I think this is the option I will go for. I would prefer 2A rather than 120V, It would be safer to use, and more useful! I like the idea that another poster suggested of just two identical supplies that I can switch to parallel or series tied, or series split, and get either +35v 4A, +70V 2A or the standard +-35V 2A. I can do this with my transformer, if I understand correctly, because of the four output wires on the secondary side. It would be really great (and I think, worth the extra cost), to have dual everything, so 2 voltage meters, 2 current meters, dual voltage knobs and dual current knobs. It would really be like having 2 bench supplies side by side in the same box.
Now even though the transformer was only $1, the rest of the parts including a nice strong steel case strong enough to hold the transformer will almost certainly come to over $100. Is that OK? If you do a good job, then an equivalent commercial linear supply could easily be over $1000. Also forget about how simple you can make a power supply with a LM317. By the time you have added everything else you need, along with all the wiring, it is a big project. I actually don't think you will want to touch a LM317. You will want a power supply which works perfectly down to zero volts and zero amps, and that has separate voltage sense wires that go to the output for best regulation. If you are smart, you can do all that with less design effort then a LM317-based supply.
That sounds fine to me, in fact if I go for four of the 10-turn pots as mentioned above, I think they alone would come to over $100. I guess I will try to start with an ideal design (aim high), in order to work out a parts list and price. Then, if I have to compromise on some things, then so be it.
If this sounds like what you are thinking about, I can suggest some next steps.
Please do, sounds great!
electronwaster
-
If it has two separate secondaries, I would probably just make two isolated positive power supply sections that you can put in series if necessary. If it's a split secondary, then it's not as easy. Designing a negative regulator doesn't seem worth the effort, it takes extra design time and increases the BOM size.
This idea really appeals to me, as I wrote a minute ago in another reply here, I would even like a full set of dual controls and meters for V and I, which I realise basically doubles the cost, but it would be really flexible!
I would arrange for a negative supply to the regulators/opamps. It doesn't seem worth to choose exotic parts like in Dave's design, you can just use a small transformer to supply the negative voltage or wrap some extra windings around the toroidal transformer. Note that if you use an external pass transistor with a three terminal regulator, you will lose the over current and overheating protection of the regulator, so make sure to take care of that.
I have lots of small transformers kicking around, I guess it only has to be small as it's not providing nearly the same power, the current could be high, but the voltage will only be around 2v or so.
Putting diodes in series will increase the output impedance and degrade load regulation, especially at low voltages. For example with a single 1N4007, at 1A the forward voltage will be about 0.9V typical, and at 10 mA it would be 0.7V. That means that 0.5 V at 10 mA will suddenly be 0.9 V at 1A. That's some pretty horrible load regulation.
I see - I hadn't thought about any of that, but it does make sense. A negative supply does sound like a good option after all.
electronwaster
-
Note that if you use an external pass transistor with a three terminal regulator, you will lose the over current and overheating protection of the regulator
isnt that possible with power resistors division to divide the current properly so that at some current fraction into the regulator, will trip the protection? say with pathetical example 0.9ohm in series with regulator and 0.1ohm at pass transistor, the regulator will trip say when 1A goes through it, while during the time 9A goes through the pass transistor, so cant we say we have 10A circuit protection now?
This is an interesting idea, but if there are better ways than using an LM317, such as opamps or comparators with big transistors or whatever, then I am not against using them at all.
For example with a single 1N4007, at 1A the forward voltage will be about 0.9V typical, and at 10 mA it would be 0.7V. That means that 0.5 V at 10 mA will suddenly be 0.9 V at 1A. That's some pretty horrible load regulation
cant we use the sense/feedback wire as amspire said to avoid this?
voltage sense wires that go to the output for best regulation
So - I may be right in that it is possible to use an LM317 down to 0V without a negative supply? To be honest that surprises me (!) as everything I read says you need the negative supply. Well - if it's true, that would or could be useful.
electronwaster
-
The LM317 needs a negative supply to voltage regulate down to 0v. It will current limit safely down to 0V without a negative supply.
If you use a LM317 as the regulator, there is no way to get separate voltage sense leads going to the output. A circuit with separate sense leads has a lot of advantages, in that you can put the current sense resistor after the regulator and so you are not measuring the current powering the regulator as well as the output current.
My idea is to start with a fully working and production tested design and then modify it for your needs. I would look at the old HP supply designs and find one with the performance you want. HP tend to have a separate supply to power each regulator circuit that is totally independent from the main supply. This allows then to have the reference circuit fed from a +/- supply that has the positive output as the regulator 0V, and there are benefits to this. To get the separate supplies, you just need to add some very cheap extra mains transformers with a center tapped output. rockby.com.au has some for just a few dollars. On advantage of this arrangement is you can have the regulator circuit fully powered up when you are debugging, and then slowly wind up the voltage to the main power transistors.
The old HP designs usually use pretty common parts, and all the small details are fully worked out - like the master/slave voltage control when the two isolated supplies are connected as a +/- dual tracking supply.
More complicated then a basic LM317 supply, but you will learn a lot more (especially since the HP manuals have a full description of how the circuit works) and you end up with a professional lab grade supply.
If you want, you can build the supply in a modular way, so you start off with potentiometers controlling voltage and current, and later you can change it to a digital control - perhaps even the digital part of Dave's circuit.
If you need any help finding a HP model to start with, I can make some suggestions.
If you want to go with a LM317-type supply, then put the big transformer aside for now, and buy a $20 transformer. Make a very compact 0-20V supply with no current limit control. It is more of a weekend project and is perfectly useable. Still have a negative supply so you can adjust down to 0 volts. I am winding my supplies down to near 0 Volts all the time, and I would just never even bother building one that couldn't work perfectly below 1.25 volts. Say you build a circuit to run off a single NiMH AAA cell battery. You need a power supply that can emulate the battery over its full range of charge, and you also want to see how the circuit behaves when the battery is flat. The single battery switching regulators still operate fine down to less then 0.7v, so you would need a supply to go down to 0.5V at least.
Richard.
-
isnt that possible with power resistors division to divide the current properly so that at some current fraction into the regulator, will trip the protection? say with pathetical example 0.9ohm in series with regulator and 0.1ohm at pass transistor, the regulator will trip say when 1A goes through it, while during the time 9A goes through the pass transistor, so cant we say we have 10A circuit protection now?
A similar circuit is described in I believe one of the LM317 datasheets.
I have lots of small transformers kicking around, I guess it only has to be small as it's not providing nearly the same power, the current could be high, but the voltage will only be around 2v or so.
In the case of three terminal regulators, you only need to provide a negative voltage for the adjust pin. Current through it will be a few mA or so. In the case of op-amps, you just need to power the op-amps from a negative voltage compared to the negative output terminal, which will also draw just a small amount of current. Hence my statement that you could probably put a bunch of extra windings around the transformer (only works for toroidal ones). This allows the regulator/op-amp to operate away from their supply rails, which makes life easier. In the HP supplies Richard references, they have separate floating supplies for this purpose. These only power the op-amps and sink base current, so current requirements are also fairly modest.
I agree with all of Richard's ideas. Especially the ones about building a simple power supply first and about using a proven design with op-amps instead of messing around with LM317s and trying to force them into being a lab supply. In a standard LM317 lab supply, people put two in series, one for current, one for voltage. That's going to cost you a lot of pass transistors at high current levels. Sure, you can add your own feedback with an op-amp, but at that point you're basically using the LM317 as a fancy power transistor (limited to 1A, although I believe there are 3A and 5A variants like the LM350 and LM338).
-
How dumb am I!
You have a toroid transformer. As long as the center hole has not been blocked off, you can add as many extra secondary windings as you like!
Just do a test. Wind 10 turns of wire around the toroid and power the primary with 240 V. Check the voltage from the 10 turns. Once you have got that number, you can easily calculate the number of turns you need to get any voltages you want. Since you have this great big transformer, you should have plenty of room for windings.
If you are going to add some digital stuff, you may need an extra supply for the digital part so it can be totally isolated from the two supplies. Communication with ADC's, DAC's, etc would be via optocouplers. This would mean that the digital circuit could be connected directly via USB to a computer, while the two supplies are completely floating.
Richard.
-
You have a toroid transformer. As long as the center hole has not been blocked off, you can add as many extra secondary windings as you like!
Just do a test. Wind 10 turns of wire around the toroid and power the primary with 240 V. Check the voltage from the 10 turns. Once you have got that number, you can easily calculate the number of turns you need to get any voltages you want. Since you have this great big transformer, you should have plenty of room for windings.
Richard, this is interesting, the extra windings for the low current isolated source, is it only applicable to toroid ? How bout EI core type ?
Also about measuring the 10 turns, is that using a direct AC volt measurement ? How to proceed next (eg. formula or calculation) once we got that 10 turns winding voltage ? Say we want to make this extra low current source with a center tapped for supplying +15 and -15 volt ?
-
You can measure the 10 turns with your multimeter on AC. Say it reads 5V (it won't, but let's pretend for a moment), then you know you need 30 turns to get 15 volts. Required turns = 10 * desired_voltage / measured_voltage
In this case, two of these windings in series gives 30V or -15:0:+15 if you reference from the centre "tap".
-
Richard, this is interesting, the extra windings for the low current isolated source, is it only applicable to toroid ? How bout EI core type ?
It's not very practical with an EI core, since it's hard to get the wire around the 'E' without taking the thing apart and removing tape/housing.
Also about measuring the 10 turns, is that using a direct AC volt measurement ? How to proceed next (eg. formula or calculation) once we got that 10 turns winding voltage ? Say we want to make this extra low current source with a center tapped for supplying +15 and -15 volt ?
Just a direct ACV measurement. Voltage should be linear with the number of turns (depends on turns ratio).
-
Some EI core transformers have a slight gap around the side of the secondary, and you can slide some extra thin wire turns on, but usually the designers fill up all the available bobbin space.
It is possible to pull apart the EI core transformers (as long as they are not welded) but as transformers are varnished, it can be a lot of work breaking the varnish apart without bending or damaging the laminations. If you can take it apart, you can totally rewind the secondary any way you like. It definitely can be done - I actually have a power supply I built in the 70's still with a rewound transformer.
I tend to only attempt to rewind transformers that have the isolated primary and secondary sections on the bobbin. If both windings are in the same bobbin section, it is probably best to leave that transformer.
Richard.
-
Okay - I think I'm starting to understand that an LM317 will not really help me much. Once I've worked around the shortcomings: power limit, two lots of volt drop, i.e. the benefits are outweighed by the problems. I may make two supplies: Firstly, what you suggested, to start with an easy, very basic, power supply, and there are a million and one schematics available for that. I think I can live with mostly any reasonable diagram, then soon after, the toroid can then be used as a dual supply, larger voltage, larger current supply.
I love the idea of starting with an HP design and customising it to my taste. I haven't googled for it yet, but I'm assuming the old schematics are widely available? If you can point to a model that most closely matches my feature requirements (or preferences, really), then I'd be more than grateful:
- dual 0-30V (or so) 2A supply - able to be used separately, in series or in parallel
- linear
- voltage control (and readout)
- current limit (and readout)
- short circuit protection
- if fan cooled, then auto variable-speed, or at least auto on-off
- (anything missing?)
As far as digital interfacing goes, I really don't think I'll need that any time soon (or ever, even). Same with dual tracking voltages. I think it would be more flexible, and no big deal to just adjust both controls to match. Unless I am missing something.
I don't think I'll need to rewind any transformers, will I still need to add extra windings to get a negative supply for running the op-amp based circuits (or whatever you have in mind)?
electronwaster
-
if fan cooled, then auto variable-speed, or at least auto on-off
I hit the jackpot with an ATX power supply I disassembled. There is a little 12 V fan speed control daughter board with a remote temperature sensor attached to a heat sink and an output socket for the fan. I'm definitely going to use it in something I build.
-
To electrode , alm and Richard, thanks for the winding info.
About adding the extra winding, I will not tear down the EI plates since the transformer is perfectly coated, looks like it was dipped completely in the lamination liquid for finishing. As Richard mentioned, there is a small gap left at the bobbin and I think I could squeeze in thin copper enameled wire in there.
@electronwaster
Check this out the Agilent/HP linear power supply like these here, it has the circuit that is using common components :
http://128.238.9.201/~kurt/manuals/manuals/HP%20Agilent/HP%20E3620A%20Operating%20&%20Service.pdf (http://128.238.9.201/~kurt/manuals/manuals/HP%20Agilent/HP%20E3620A%20Operating%20&%20Service.pdf)
or this
http://128.238.9.201/~kurt/manuals/manuals/HP%20Agilent/HP%20E3610A,%2011A,%2012A%20Operating%20&%20Service.pdf (http://128.238.9.201/~kurt/manuals/manuals/HP%20Agilent/HP%20E3610A,%2011A,%2012A%20Operating%20&%20Service.pdf)
-
Check this out the Agilent/HP linear power supply like these here ...
Thanks for those links, I will check those out right now.
Thanks to the other poster too with the idea about the fan speed sensor. I have a couple of old, dead PCs in the workshop at the moment, and I might take a look inside the power supplies.
electronwaster
-
I prefer fanless power supplies. That is my preference but fanless is quieter and more reliable. Well built linear supplies from the 60's and 70's often still work perfectly now, which is why I like to see the design made from long life parts only. A supply I made about 1972 was upgraded once with better opamps a decade later, is still a great supply today.
It is nice not needing an over-temperature shutdown that you would need for a fan-cooling based supply.
Concerning the output voltages that you can get from a winding, you have to take into account the minimum mains voltage, and the amount of ripple on the capacitor. It is not always good having too large a capacitor, as the bigger the capacitor, the higher the peak current from the transformer. The higher the peak current, the more likely the core will momentarily saturate which will make the transformer hotter, and will probably make the transformer buzz annoyingly.
So with a winding of 25V AC, I would reduce that to about 22V for low mains. Rectified peak, you get about 30 volts peak. Allowing for 2V P-P ripple, and 1volt drop across the regulator transistor and the current sense resistor, you get about 27 volts maximum DC output from the power supply. With a voltage doubler rectifier from the transformer instead, you could increase the output from each supply to 50V DC. With the two in supplies series you would have 100 volts.
I like to allow for a capacitor to be able to decrease in value with age without affecting the power supply regulation. So if I design for 1V p-p ripple, I would allow for a 2V p-p lifetime ripple.
Richard.
-
With a voltage doubler rectifier from the transformer
Would this be a voltage doubling rectifier?
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20280;image)
-
Another thing you can do is make the doubler switchable, if for example you want double voltage, but a lower current capacity, and vice-versa:
(http://upload.wikimedia.org/wikipedia/commons/8/86/SwitcheableRectifier.jpg)
Bear in mind that your ripple considerations will change with the configuration though...
-
Would this be a voltage doubling rectifier?
Yes. The ripple across the capacitors is at 50Hz instead of 100Hz for a full wave rectifier, and also the two capacitors are in series so that the ripple gets added. So the end result is the ripple for, say, two 1000uF capacitors in the doubler circuit is 4 times the ripple in a single capacitor in a full wave rectifier circuit. So you either use bigger capacitors, or you live with greater ripple.
To get 5Vp-p ripple with 1A out, you would need 2x 10,000uF 35VW capacitors in the doubler circuit. In a full-wave circuit, the same two capacitors in parallel would give 1.2V ripple at 2A output.
About the power supplies, look at something like this: http://128.238.9.201/~kurt/manuals/manuals/HP%20Agilent/HP%20E3632A%20Service.pdf (http://128.238.9.201/~kurt/manuals/manuals/HP%20Agilent/HP%20E3632A%20Service.pdf)
It is a single supply, and the whole digital side can be hugely simplified with today's parts. There is also part of the circuit you can cut out, like the swithes on the different power supply taps to change the voltage. The circuit is based on the very common TL074 opamps.
Also, to see how dual supplies are interconnected, you can look at this:
http://128.238.9.201/~kurt/manuals/manuals/HP%20Agilent/HP%206227B%20Operating%20&%20Service.pdf (http://128.238.9.201/~kurt/manuals/manuals/HP%20Agilent/HP%206227B%20Operating%20&%20Service.pdf)
The 6227B is a pretty good supply as well, and you would not need the two sets of power transistors that are there to change voltage, as you will only have one rectified voltage from your transformer.
Richard.
-
Another thing you can do is make the doubler switchable, if for example you want double voltage, but a lower current capacity, and vice-versa:
That is an interesting idea.
electrode is right. In the full wave mode (switch off), you can double the current to get the same ripple as in the voltage doubler mode (switch on).
So using my numbers from my last post, at 1A output, two 10,000uF capacitors will have just under 5V p-p ripple in the doubler mode. In the full wave mode, the ripple would have to under 2.5V at 1A output or 5V p-p at 2A output.
Switcheable voltages are useful for halving the maximum heatsink power you need to get rid of. You have some kind of circuit that turns off the switch whenever the output voltage is below about 22-24 volts. The rest of the time it is on.
-
Should also mention that I got that image from Wikipedia - it wasn't some new idea of mine.
Thanks for posting the power supply pdf links btw.
-
About those HP linear supply circuit, looks like they're all using separated auxiliary windings for the digital metering circuit at the front panel and the voltage reference.
For the measurement circuit , its easy to understand, but for the voltage reference, any reason why it needs a separate winding ? Cause I've seen a lot of linear type adjustable psu, most of the V-Ref are getting it's power from the main power winding.
PS :
@electronwaster : Don't mean to hijack this thread of yours, its just I have the same interest in building linear psu too, if this is a problem, just say so.
Below attached pic is my transformer that has been gathering dust for years, it has the auxiliary winding 18,12,ct,12,18 below the primary winding terminations.
-
PS :
@electronwaster : Don't mean to hijack this thread of yours, its just I have the same interest in building linear psu too, if this is a problem, just say so.
No worries, I'm learning as we go! :-)
-
BravoV, that is an impressive transformer. Over 600VA with two separate multi-tapped windings.
Must be heavy.
-
With a voltage doubler rectifier from the transformer
Would this be a voltage doubling rectifier?
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20280;image)
currently working on this, tried to figure out how to get +ve and -ve rail from a 2 pin output transformer, it turned out what i actually got is a Delon Bridge Voltage Doubler (above circuit). since its infront of me, let me turn it on and capture some photos for entertainment. the transformer is 2 pin output giving 20Vpeak (40Vpp) output. i use my 12V 50W video light to test load the bridge. i used 35V 3300uF smoothing caps on both sides. here's the image:
1) i get +20V and -20V rail, can be configured as +40V output (voltage doubler):
2) loading the +ve rail (50Hz ripple, bad attenuation)
3) loading the -ve rail (50Hz ripple, bad attenuation)
4) loading the doubled voltage, ie +ve to -ve rail (100Hz ripple (refer to purple channel = yellow - blue potential), better attenuation, similar to full bridge effect)
-
Hi there,
I bought an altronics branded 25-0-25 300VA toroidal transformer from the local tip shop (recycling centre) for $1. It appears unused as the tips of the leads are still enamelled - I'd like to build a bench power supply with it. I'm currently using (playing around with) a modified antec PSU with multiple outputs but no voltage adjust or current limit (except the built in thermal overload or whatever), and I would really like these features, so I think I may use the transformer as a basis for a nice linear bench power supply.
you had a bargain there mate.these recycling centres.why don't we have these in the uk .it's crazy.
-
This is an alternative doubler circuit.
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20296)
Here is a puzzler. This circuit has a single capacitor across the rectified output. The other doubler circuit has two in series which halves the capacitance, and so doubles the ripple if you were using the same capacitors. So I can use 5,000uF capacitors instead of 10,000uF capacitors for the same ripple. That sounds twice as good.
But this circuit will probably end up being inferior to the other doubler circuit as a means of providing the voltage doubled output. Why?
Richard.
-
HP AN-90B would also be a good read. It discusses many of the issues with designing a linear power supply. I don't have a link handy, but I'm sure your favorite search engine can track it down in a few seconds.
About those HP linear supply circuit, looks like they're all using separated auxiliary windings for the digital metering circuit at the front panel and the voltage reference.
For the measurement circuit , its easy to understand, but for the voltage reference, any reason why it needs a separate winding ? Cause I've seen a lot of linear type adjustable psu, most of the V-Ref are getting it's power from the main power winding.
You need a supply a few volts below the negative output terminal (for 0V operation) and one a few volts above the max. output voltage (eg. for driving the base of the pass transistor) anyway. You could connect the positive output of your 'negative' supply to the negative output terminal, and use something like a peak detector for the positive supply, but that means that the voltage across your reference will vary wildly. It also requires the op-amps to handle the full output voltage + negative supply + positive supply, which might be something like 40 V for a 30 V supply. Stability will probably also be better when referencing the op-amp rails to the output voltage.
-
Yes. The ripple across the capacitors is at 50Hz instead of 100Hz for a full wave rectifier, and also the two capacitors are in series so that the ripple gets added. So the end result is the ripple for, say, two 1000uF capacitors in the doubler circuit is 4 times the ripple in a single capacitor in a full wave rectifier circuit. So you either use bigger capacitors, or you live with greater ripple.
I looked into this as the argument above did not seem intuitive to me.
First of all I simulated a full wave bridge configuration on a nominal 12 V 60 Hz AC source with a 1 A load presumed to come from a downstream voltage regulator. Smoothing is provided by two 5000 uF capacitors paralleled up:
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20318;image)
As can be seen the ripple is 657 mV on top of 15.4 V minimum sustained voltage. This is about 4.2%.
Next I reconfigured the system as a voltage doubler using the same two capacitors. The load current is halved in light of the doubled voltage to preserve a similar power output:
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20320;image)
Here the ripple is 1.30 V on top of a 31.1 V base. Once more this is about 4.2%, essentially the same in relative terms.
This is intuitive, since we have the same size energy reservoir in the system in each case (the same two capacitors), and there are essentially no energy losses save for the rectifying diodes. I think we have to accept a higher absolute ripple value on the higher output voltage just from considerations of proportionality.
-
IanB,
That all looks right. Nice job. Since you have done the modelling, it is worth plotting AC current with different capacitor sizes. Try a 100,000uF capacitor.
Richard.
-
Try a 100,000uF capacitor.
Hey, do you know how much those things cost? ;D
But since you suggest it, that is an interesting experiment.
Here is the bridge rectifier circuit with 10,000 uF. I used a 0.5 ohm resistance for the transformer secondary winding based on the measured value of a real 12 V 2 A transformer I have:
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20324;image)
And then with 100,000 uF:
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20326;image)
The ripple is very nicely reduced.
I think you might have been expecting larger surge currents from the transformer with the bigger capacitor? Apparently it didn't happen that way.
-
I was expecting a higher peak current, but since you have the 0.2 ohm resistor in, that limits it.
If you did a startup simulation, you would see a big difference.
-
BravoV, that is an impressive transformer. Over 600VA with two separate multi-tapped windings.
Must be heavy.
Yeah, I think its close to 10 Kg and can't imagine to drop this thing on your toe accidentally. ;D
HP AN-90B would also be a good read. It discusses many of the issues with designing a linear power supply. I don't have a link handy, but I'm sure your favorite search engine can track it down in a few seconds.
Thanks, downloaded and on a quick glimpse it does impress me, definitely a keeper.
You need a supply a few volts below the negative output terminal (for 0V operation) and one a few volts above the max. output voltage (eg. for driving the base of the pass transistor) anyway. You could connect the positive output of your 'negative' supply to the negative output terminal, and use something like a peak detector for the positive supply, but that means that the voltage across your reference will vary wildly. It also requires the op-amps to handle the full output voltage + negative supply + positive supply, which might be something like 40 V for a 30 V supply. Stability will probably also be better when referencing the op-amp rails to the output voltage.
Ok, thanks again for the explanation, btw what is that "peak detector" are you talking about ?
-
I was expecting a higher peak current, but since you have the 0.2 ohm resistor in, that limits it.
If you did a startup simulation, you would see a big difference.
I put the transformer secondary resistance in because I wanted the simulation to be more true to life. It is the engineer in me, I'm afraid.
For comparison, here are the same two cases with no source resistance:
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20328;image)
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20330;image)
The startup does indeed show some significant current surges. I put the source resistance back in this time otherwise the simulation results become quite unrealistic:
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20332;image)
In this case the current peaks reach a magnitude that could possibly exceed the maximum instantaneous current ratings of the diodes. That is something I will need to take note of.
-
Interested with IanB simulation, especially at the 100000uF ??? filter cap, added a few extra discrete components to limit the inrush current. All extra components for the inrush limiter are picked from the ltspice library, nothing special.
Question, will this work in the real circuit as in the simulation ? Did I miss anything or it can be built with much more simpler circuit ?
Bottom trace is the voltage difference between Gate and Source pin at the P-mos.
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20339;image)
-
Ok, thanks again for the explanation, btw what is that "peak detector" are you talking about ?
Just a diode and a capacitor connected to the AC signal (or rectified signal). As long as you barely load the output, the voltage will be very close to one diode drop below the peak of the input voltage, while the emitter voltage of the pass transistor will be lower due to ripple, loading and finite buffer capacitance (as long as you don't use an insane amount of capacitance ;)).
-
Since there has been some discussion about regulating down to zero volts and needing a negative supply voltage with many regulators or op amps to achieve this, I thought I'd post an illustration of one way to get a low current negative supply without needing any extra windings on the transformer. The negative supply obviously needs filtering before it can be used, but something like a 7905 negative regulator should be OK for that.
(https://www.eevblog.com/forum/beginners/linear-power-supply-ideas/?action=dlattach;attach=20355;image)
-
Interested with IanB simulation, especially at the 100000uF filter cap, added a few extra discrete components to limit the inrush current. All extra components for the inrush limiter are picked from the ltspice library, nothing special.
Question, will this work in the real circuit as in the simulation ? Did I miss anything or it can be built with much more simpler circuit ?
It looks like you put a current limiting circuit in there. However, on looking at the datasheet for a power diode, I saw that it was rated for a momentary current pulse many times larger than the maximum average current. I don't recall the number but it was in the order of 600 A! So I would think that currents up to 50 A per mains cycle would not be an issue for a typical power rectifier and my concern was over pessimistic. As always, check the datasheet for the particular device you are planning to use...