EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Kane on April 28, 2023, 03:03:28 pm
-
Hi,
I'm looking to build a linear power supply as a learning process and I have a 24V transformer that outputs 28.45VAC unloaded and around 38.75VDC after the bridge rectifier at the input filter cap.
This voltage is too high for the LM324N I was going to experiment with.
How do I choose what transformer to use if I say wanted a regulated up to 24V 1A supply?
Should I get a different transformer or would I just use a Linear regulator like an LM317 to reduce the voltage powering the LM324N to say 30V?
What is a good way of dealing with this?
-
This is always the issue with linear supplies. Unless your transformer is way to big for the load, the voltage is going to drop under actual usage. Likewise, the capacitor after the rectifier measures that peak value with no load, but there will be ripple as you add load. Under full load--whatever the transformer is rated for--the two effects will cause the lowest voltage points across that capacitor to drop near 24V. A 30V output from an LM317 requires that you maintain its input voltage at ~32V or more. That would only work if the transformer and capacitor were oversized for a given load.
You need some actual numbers. What current output do you need from the supply? What specific transformer, rectifier and capacitor are you using?
-
No experience with op-amps, but I would guess they preferably would
want a stable voltage. A transformer plus rectifier and capacitor will
not give a stable voltage, so I'd guess you need some kind of regulation.
A rule of thumb is to use ~2000uF/A and that it will give you around
4V ripple. And a LM317 (or 78xx) needs 2-3V headroom to operate, so
the transformer will at least be at 30V peak after rectification (at 1A).
-
This is always the issue with linear supplies. Unless your transformer is way to big for the load, the voltage is going to drop under actual usage. Likewise, the capacitor after the rectifier measures that peak value with no load, but there will be ripple as you add load. Under full load--whatever the transformer is rated for--the two effects will cause the lowest voltage points across that capacitor to drop near 24V. A 30V output from an LM317 requires that you maintain its input voltage at ~32V or more. That would only work if the transformer and capacitor were oversized for a given load.
You need some actual numbers. What current output do you need from the supply? What specific transformer, rectifier and capacitor are you using?
The transformer is: https://uk.farnell.com/multicomp/mcta080-12/transformer-toroidal-2-x-12v-80va/dp/9532706 (https://uk.farnell.com/multicomp/mcta080-12/transformer-toroidal-2-x-12v-80va/dp/9532706)
The rectifier is: https://www.mouser.co.uk/ProductDetail/821-TS20K100-T (https://www.mouser.co.uk/ProductDetail/821-TS20K100-T)
The capacitors are some 63V 4700uF Hyncdz brand ones I got off eBay
My goal is primarily learning about CV & CC concepts and digital control but I would like to make a low noise supply around 24V 1A max because I only have a nasty cheap switching supply at the moment.
-
My goal is primarily learning about CV & CC concepts and digital control but I would like to make a low noise supply around 24V 1A max because I only have a nasty cheap switching supply at the moment.
OK, the transformer and rectifier seem appropriate enough for that, plenty large but not gross overkill. The RMS AC current from the transformer will be larger than the average DC current output by as much as a factor of two, so you have a good safety margin but not 3.33X as you might think.
Using an LM317 to lower the supply voltage to something like 28-30V is a good start, but you are going to quickly run into thermal issues--linear PSUs regulate by dissipation. If you want a variable PSU with separately adjustable CC and CV controls, I'd suggest you design it from the start with a 2-level input where for voltages less than 12V, you switch your transformer to the 12VAC tap and lower your preregulator to 14-15V. If you are using digital controls, this should be easily accomplished.
-
My goal is primarily learning about CV & CC concepts and digital control but I would like to make a low noise supply around 24V 1A max because I only have a nasty cheap switching supply at the moment.
OK, the transformer and rectifier seem appropriate enough for that, plenty large but not gross overkill. The RMS AC current from the transformer will be larger than the average DC current output by as much as a factor of two, so you have a good safety margin but not 3.33X as you might think.
Using an LM317 to lower the supply voltage to something like 28-30V is a good start, but you are going to quickly run into thermal issues--linear PSUs regulate by dissipation. If you want a variable PSU with separately adjustable CC and CV controls, I'd suggest you design it from the start with a 2-level input where for voltages less than 12V, you switch your transformer to the 12VAC tap and lower your preregulator to 14-15V. If you are using digital controls, this should be easily accomplished.
Thanks for the reply,
I'm aware of the heat dissipation relating to linear supplies, I just wanted to start off simple say a LM317 and a 10 turn pot and work my way through and explore other concepts to learn. I may try switching transformer taps.
Another thing I was looking at trying was a buck converter that sits 3V above the output for the dropout of a linear regulator. Is this something worth looking into?
I'm not sure what to realistically expect relating to noise levels and if it will be possible to filter out most of the switching noise for an efficient low noise supply.
-
Hey Kane
Me too building a linear power supply right now. I am using a pass transistor regulation circuit.
I am using 16x2 display for displaying current and voltage.
-
The transformer is not so bad. Instead of another transformer one could choose a different OP-amp (e.g. the MC33174, that is relatively close to an LM324 but can withstand up to some 44 V supply).
There are also circuits with an extra voltage regulation for the OP-amps. So the OP amps could run on some limited 24 V or so.
With only 1 A of maximum current one could get away without tap switching, though for learning one could still include it.
A relatively common Lab supply circuit use a seprate transformer for the regualtor and possible digital control / display. So one could still use the existing transformer for the main power and only add a small one (e.g. 2x10V 5 VA) for the regulator.
-
Another approach is to use a floating regulator circuit, with its control circuit's supply rails referenced to the positive output terminal, sensing its output voltage with a 'upside down' potential divider between +out and -out.
As yours is a toroidal transformer you could simply add an overwind to add a low current, lower voltage secondary to power it, one end tied to +out, feeding positive and negative half-wave rectifiers and linear regulators to supply the control circuit OPAMPs, which do not need to be high voltage types. Choose your auxiliary rail voltages then start by adding a temporary 10 turn winding using hookup wire, and measure the RMS voltage to get an idea of how many turns the overwind secondary will need. N.B. VA drawn from the overwind subtract from the total VA available from the original secondaries.
-
Another thing I was looking at trying was a buck converter that sits 3V above the output for the dropout of a linear regulator. Is this something worth looking into?
I'm not sure what to realistically expect relating to noise levels and if it will be possible to filter out most of the switching noise for an efficient low noise supply.
No, that would defeat the purpose of making a linear supply in the first place. The buck converter will generate switching noise and linear regulators aren't good at suppressing that. Other methods of filtering will be cumbersome and probably not nearly as effective as just sticking with a linear design.
-
I took a power supply circuit out of a control box and replaced it with an off the shelf regulated power supply. The circuit includes a transformer that step down to 19VAC. A full wave rectifier bridge and a 20,000 microfarad capacitor. It delivered about 25VDC no load and 23VDC at full 5A load. I think it's quite good for an unregulated power supply.
-
Do you really need 30V for these experiments?
the LM317 has a fairly high dropout voltage. To get 30V out of it you will need a minimum 33V or better going in to it to guarantee decent ripple rejection. I don't see your current collection of parts managing that at 1A. You might be able to get a nice stable 24V with enough heatsinking.
-
There's old versions out there with just BJT's, that are fairly adaptable to different transformers. Always do some basic power calculations of the different parts at the lower and upper limits of operation, and make a model in a simulator if u can. I burned a few zener diodes on my homemade BJT PSU, before I changed a few things. But it works pretty good, I left it to run at ~1/2 output power overnight, and the DC average drift on my 6.5d DMM, was only like 5-10mV, or even 2-3mV. It wasn't much that's for sure.
That one needs 2-3 windings tho.
-
No experience with op-amps, but I would guess they preferably would
want a stable voltage. A transformer plus rectifier and capacitor will
not give a stable voltage, so I'd guess you need some kind of regulation.
Op amps are actually very tolerant of supply voltage variation, it is or at least was fairly common to power them with unregulated supplies.
-
Even the old lm317 has decent ripple rejection in the 100khz-1mhz range, for the switching noise you would need some extra inducter/capcitors. A lt3080 would probably be better for following a switching supply though and it's dropout voltage is under 0.5v @ 1amp.
-
Instead of a regulator, I would tend to use a zener diode and maybe pass transistor to make a slightly lower regulated supply for the LM324.
I never really faced this specific problem because I prefer single operational amplifiers in this sort of application, and their supply voltage is often 44 volts instead of 32 volts.
Another approach is to use a floating regulator circuit, with its control circuit's supply rails referenced to the positive output terminal, sensing its output voltage with a 'upside down' potential divider between +out and -out.
Quad and dual operational amplifiers are less amenable to floating regulators, but that could work.
-
Hi,
I'm looking to build a linear power supply as a learning process and I have a 24V transformer that outputs 28.45VAC unloaded and around 38.75VDC after the bridge rectifier at the input filter cap.
This voltage is too high for the LM324N I was going to experiment with.
How do I choose what transformer to use if I say wanted a regulated up to 24V 1A supply?
Should I get a different transformer or would I just use a Linear regulator like an LM317 to reduce the voltage powering the LM324N to say 30V?
What is a good way of dealing with this?
Hi,
The LM324 will work at voltages much less than 24 volts. Even 5v would power it for experiments.
-
The 24 V transformer is actually not that bad to get about 25 V out. Under load the AC voltage will go down to about the nominal 24 V and so will the the DC peak voltage. In addition there are some 10-20% lost to the ripple voltage. So the voltage with ripple maydrop to some 27-29 V depedning on the capacitor size. Depending on the power stage 2-4 V drop is not that bad - it is even on the low side, so that one has to take some case no to loose too much. E.g. the LM324 driving a darlington emitter follower with plenty of emitter resistance for load sharing and a relatively large shunt may even be on the high side and need an extra voltage for then LM324.
-
Your transformer is 2x12V (4 wires), right?
You may build yourself the PSU with 12V transformer output, and on top of that switch in the additional 12V segment when required.. Each 12V segment consist of its own diode bridge and filter capacitor.
The electronics powered from the first 12V segment, and the pass transistor(s) powered either from the first 12V segment (like 1V..12V regulated output), or from first+second (like 12V..24V regulated output)..
-
You want to learn about OP amp circuits. OK.
Rule #1: OP amps are DESIGNED to work with dual power supplies. That means equal positive and negative Voltages. Sure, they can be operated on single supplies, but that is the hard way. Why, you ask. Because the inputs and output are all going to be around the half way point between the positive and negative supply rails. So, with a single supply around 30 Volts, your ins and outs are going to be around 15 Volts. And if one rail (the positive one) has some ripple (because it is not regulated) while the other rail (ground in the case of a single supply) has none, then that mid point is going to be jumping up and down at the ripple frequency. And your inputs and outputs will be jumping with it.
Can you work around this? Perhaps. But I have designed a number of OP amp circuits and I would NOT want to add that complication to the design unless it was absolutely necessary.
So my humble suggestion is two fold: first use a dual Voltage supply. And second, avoid the ripple by using linear regulators in both rails of that dual supply. This will make learning a lot simpler. Then, when you are an expert, try single Voltage supplies or unregulated ones - or even both if you have a lot of time.
Your transformer has two secondary coils, each providing 12 VAC. So you can use it for a dual Voltage supply. Tie the two secondary coils in series as you would to produce 24 VAC and the common point becomes your ground. The two ends are connected to the AC inputs of your bridge rectifier and the + and - outputs of that bridge rectifier are about (12 V x 1.4) - 0.7V = 16.1 Vp-p.
Now, the ripple: The generally accepted way to calculate ripple with a simple filter capacitor is:
Vr = I/2fC
In your case, you need to assume a maximum I which you say is 1 Amp, f is probably 50 Hz, and C is 4700uF or 0.0047 F. So the ripple Voltage, Vr is going to be about 2.13 Vp-p. Subtracting that from the 16.1 V above gives us about +/-14 V to play with. By the way, if you want a smaller ripple Voltage, you can just turn that equation around and find the value of filter capacitor needed for whatever value of Vr that you want. C = I/2fVr
If you insist on using an unregulated supply, then there you have it. +/-14 VDC with between 0 and 2.13 Vp-p ripple and the ripple will be about equal and opposite on the two supply rails so the effect on the OP amps will be minimized. OP amps really do work best with dual supply Voltages.
But I would add Voltage regulator chips. They are inexpensive and work very well. Fixed + and - 12 Volt regulators would probably work: the 7812 and 7912 fixed Voltage regulators have a drop out Voltage of 2 Volts so they should work with the +/- 14 VDC unregulated Voltages. Each would dissipate 2 Watts of power so simple heat sinks would be OK. If you want adjustable Voltages, a pair of 7805 and 7905 regulators could be configured with a dual variable resistor to give tracking, adjustable Voltages from 5 to 12 Volts. Of course, the power dissipated would increase to 18 Watts at the lower Voltage setting and full 1 Amp output currents. So larger heat sinks would be needed.
The Voltage budget is close for the +/- 12 VDC outputs, but then, your transformer will probably output a bit over +/- 12 VAC so there should be a little extra in the mix. And, if it becomes critical and the transformer is not able to keep up at 1 Amp output level, you can always replace the silicone rectifier with a Shockley one or add larger filter capacitors.
Anyway, that is how I would approach it.
Oh, and you may notice that, using my power supply design method, the AC Voltage rating of the power transformer is just about equal to the DC output Voltage after the linear regulators. Funny how that works out.
-
Rule #1: OP amps are DESIGNED to work with dual power supplies.
Not all operational amplifier are designed to work with dual power supplies, just like almost all operational amplifiers are not designed to work at one supply voltage. Besides input and output voltage ranges, the specifications which cover this are common mode rejection and power supply rejection.
The input offset voltage, and other specifications, are given with one or maybe a couple of different supply voltage configurations. Variation of the input offset voltage with single and dual supply operation depends on the common mode rejection. Variation of the input offset voltage with total supply voltage depends on the power supply rejection.
For example, an OP-07 has specifications for +/-15 volts, but what can be expected when operating with a 30 volt single supply? The input offset voltage will change by up to 75 microvolts, and more likely less than 15 microvolts.
And if one rail (the positive one) has some ripple (because it is not regulated) while the other rail (ground in the case of a single supply) has none, then that mid point is going to be jumping up and down at the ripple frequency. And your inputs and outputs will be jumping with it.
Ideally the inputs and outputs would be unaffected by supply voltage variations, but there is some effect which is specified in the common mode and power supply rejection. For a part like the LM324, I would expect an input offset voltage variation of perhaps 500 microvolts per volt of supply variation *absolute worst case*, with 10 microvolts per volt being more typical. That is comparable to the line regulation of an LM317. This variation is actually pretty high as the LM324 is *not* a precision part. If you want better performance, then regulate the supply or use a better operational amplifier.
Can you work around this? Perhaps. But I have designed a number of OP amp circuits and I would NOT want to add that complication to the design unless it was absolutely necessary.
Tektronix liked to use standard operational amplifiers with asymmetrical supply voltages to allow operation down to ground, or slightly negative, which has the advantage of preserving most of the operational amplifier's output range. So for instance they had 741s or 301As running on a supply of -5 volts and +27 volts to control their 20 volt output power supplies, leaving just enough output range to control a Darlington or Sziklai pair.
Why didn't they use a single supply part, other than lack of availability? They deliberately offset the operational amplifier negative, using the negative supply, so that with zero input the output was slightly negative, which guaranties that the output of the power supply can reach zero volts output.
If they had used symmetrical supplies, then control of the output transistor would have required the extra complexity of an additional gain stage within the feedback loop.
-
Back to transformer selection:
For traditional 50/60 Hz transformer-rectifier-filter power supplies, a useful freeware is PSUD2 from Duncan Labs.
https://www.duncanamps.com/psud2/ (https://www.duncanamps.com/psud2/)
Note that the transformer is specified by its no-load rms voltage and equivalent resistance at the secondary.
Manufacturers usually specify transformers by secondary voltage at a rated current, which is lower than the no-load voltage.
There are suggestions about how to estimate the no-load voltage and resistance, but you can measure them easily.
It is useful to see the start-up behavior, and to estimate the ripple currents in the filter capacitors, as well as estimating the final ripple voltage as a function of current.
There is a short list of rectifier models, including PN, Schottky, and vacuum-tube.
-
Back to transformer selection:
For traditional 50/60 Hz transformer-rectifier-filter power supplies, a useful freeware is PSUD2 from Duncan Labs.
https://www.duncanamps.com/psud2/ (https://www.duncanamps.com/psud2/)
Note that the transformer is specified by its no-load rms voltage and equivalent resistance at the secondary.
Manufacturers usually specify transformers by secondary voltage at a rated current, which is lower than the no-load voltage.
There are suggestions about how to estimate the no-load voltage and resistance, but you can measure them easily.
It is useful to see the start-up behavior, and to estimate the ripple currents in the filter capacitors, as well as estimating the final ripple voltage as a function of current.
There is a short list of rectifier models, including PN, Schottky, and vacuum-tube.
It seems hoops need to be jumped though to get that software. I can find no link to it at the site you posted. They say you can get the latest software at another site, but when you go there .. no links to download. I am assuming they require you to join their "group" to get the software. :palm: Yet another un-secure place to put your credentials on the internet and yet another log-in and password to remember. Phuk! Why? Just why? :-// :-// :-// :-// :-//
Not worth the effort.
-
How do I choose what transformer to use if I say wanted a regulated up to 24V 1A supply?
Hi
Consider that the regulator needs some volts between input and output, let's say 4V min.
Then, you need 28V at the input.
You have to add the diodes forward voltage: 2V for a bridge: 28+2=30V.
The ripple:
VC=IT
If I=1A and T=10ms (at 50Hz; at 60Hz ripple will be lower):
VC=10m
C=10m/V (V is the ripple)
If we accept a 2V ripple at full load (1A):
C=10m/2=5mF=5000uF (C=4700uF 40 or 50V).
30+2=32V (Vp).
Vrms=Vp/sqr(2)=32/1,41=22,7V.
We add 5-10% for transformer load and mains under voltages (today 5% should be fine):
22,7*1,05 = 23,8V.
You can use a 24V transformer.
-
Sorry about the download problems for PSUD2.
I have found it very useful, and will try to find a better download site.
-
Sorry about the download problems for PSUD2.
I have found it very useful, and will try to find a better download site.
It's not your fault Tim.
-
Sorry about the download problems for PSUD2.
I have found it very useful, and will try to find a better download site.
It's not your fault Tim.
It seems that things have changed since I downloaded it, and it is only available now through that group.
I have no experience with that group, but I still recommend the software.
-
One can use a more normal more capable simulation tool (spice variants like LTspice or Tina) also to simulate the ripple voltage part.
For a serious design of a lab supply using a simulation is anyway a good idea. For the normal small signal stability one could use pen and paper - though not that easy. The more tricky part is however the cross over between CC and CV mode and also possible large signal (e.g. including saturation at some point) oscillation. These are hard to handle the old style.
-
I believe PSUD2 is Spice-based.
It has a convenient UI and a reasonable collection of diode models.
It was free when I downloaded it--I don't know the details about this group.