-
One problem down, two new ones crop up -- prob w/ LM317/337 regulators
Posted by Praxis on 14 Aug, 2011 21:40 -
So, I'm working on a classic simple bench-top split power supply, using this schematic: http://www.aaroncake.net/circuits/supply3.asp
Well, one problem is that I'm getting about 26 volts out at either rail, when the transformer should give +/- 18v. I've got a 680 ohm resistor across the outputs as a load, and I'm measuring the output across the resistor.
Another more serious problem is that if I crank the pots down the bloody sense resistors incinerate themselves. Vout only gets to about 23 or 24v when this happens. I'm using 220 ohm 1/4W resistors.
I've checked and re-checked the connections and problems. Any ideas why this is happening? -
What are the voltages on C1 and C2?
Did you take care to connect pins 1, 2 & 3 correctly on the regulators?
Did you take care not to connect the regulator heat sinks to any electrical points they should not be connected to?
Did you wire R2 and R1 to the right points in the circuit?
What is the voltage across R2 (& R3) when they are not smoking?
Are you sure you connected R2 and R3 to the right points in the circuit? -
Well, one problem is that I'm getting about 26 volts out at either rail, when the transformer should give +/- 18v. I've got a 680 ohm resistor across the outputs as a load, and I'm measuring the output across the resistor.
18Vrms AC after full-bridge rectification will give sqrt(2)*18V DC, because the diodes + caps work as peak detectors. The DC voltage is close to the peak AC voltage, which in the case of a sine is sqrt(2) times larger than the RMS voltage that's usually quoted. Subtract about 2V for the drop across the two diodes. You can probably find some illustrated examples on a site like Wikipedia. Note that max current decreases with even more than sqrt(2), because of the uneven loading of the transformer (caps only charge during the peaks). About 0.6 is usually used.Another more serious problem is that if I crank the pots down the bloody sense resistors incinerate themselves. Vout only gets to about 23 or 24v when this happens. I'm using 220 ohm 1/4W resistors.
V^2/R = 2.6W for 24V/220ohm, no wonder that they commit suicide. The 24V output voltage is wrong, however, it should be about 1.25V if R1/R4=0 ohm. Are you sure that the wiring is correct, i.e. no short between the regulators via the heat sink (the case of the LM317 is Vout I believe, but the case of the LM337 is Vin)? Does Vout drop if you decrease R1 at all, i.e. does it work until it goes to zero and then blow up, or is Vout a constant ~24VDC? -
Stupid question, but are you sure that you have the 317 on the positive side and the 337 on the negative? Other than that, what they said. Check for shorts between the body and the heatsink, make sure the sense resistors are wired right.
Make sure the regulators are the right way around: make sure you check the diagram in the data sheet to see if it is a bottom view or a top view!
Cranking the pots "down" should make the output voltage lower not higher. -
AronCake is full of poorly designed circuits - this is one of the better ones, at least it works. It just needs a bit of tweaking for optimal performance.
26V sounds about right for an 18V transformer, Google for RMS.
R2 and R3 should be 120R otherwise the output voltage may be too high when unloaded but if you do that you should change the pots to 2.5k.
If you plug the resistor values into the formula given on the LM317 datasheet you get 27.3V but you won't be able to get the voltage as high as this. The maximum output voltage of this circuit when unloaded will be around 24V but there won't be any voltage regulation. The maximum output voltage when fully loaded will be around 15V. This means you'll have a lot of dead band i.e. when it's fully loaded 55% of the pot's adjustment won't make any difference to the output voltage and there'll be no regulation.
I'd recommend increasing the capacitor values to 3300uF, using 1k for the pots and 82R for R2 and R3; this will give a maximum output voltage of 16.5V with minimal deadband at full load. The pots should be rated to at least 250mW. -
Note that the power rating of a pot applies to the pot at max resistance, when all of the resistive material is being used. If you've set it to just 1% of its value, only about 1% of its surface is used, and the power capacity should be derated accordingly.
-
What are the voltages on C1 and C2?
26.7 and -26.1, respectively.QuoteDid you take care to connect pins 1, 2 & 3 correctly on the regulators?
Yep. Actually, I screwed it up the first time since I didn't realise that Vin and Vout were switched on the two regulators.QuoteDid you take care not to connect the regulator heat sinks to any electrical points they should not be connected to?
Yep. The heat sinks aren't electrically connected to anything in the circuit and are insulated with mica insulator gasket thingies.QuoteDid you wire R2 and R1 to the right points in the circuit?
I'm sure they're right, now at least.QuoteWhat is the voltage across R2 (& R3) when they are not smoking?
Heh, the first time I powered this up, 'smoking' wasn't the right word, 'glowing' was.
With the pots cranked all the way up to their maximum resistance, about .635 across R3 and .960 across R2. As the pot is cranked down, the voltage gets up to about 18V and then the resistors start to get really hot.QuoteAre you sure you connected R2 and R3 to the right points in the circuit?
Yep. On both I've connected terminal 1, the pot wiper, and the resistor to the same solder blob. (Yay perfboard.) The other end of the resistor on the LM317 is connected to the middle terminal. The other end of the resistor on the LM337 is connected to the third terminal. In both cases, my Vout line is on the non-sense terminal that the resistor connects to, and the other non-sense terminal is wired to Vin. The other ends of both pots are connected together and wired to ground.
I'll admit to some dyslexia, but I've been over this thing a bunch of times, rewired my mistakes, and can't figure out what the problem is. -
Stupid question, but are you sure that you have the 317 on the positive side and the 337 on the negative?
It's not a stupid question because god knows I've made mistakes like that. In this case, though, they are receiving the correct input polarities.
QuoteMake sure the regulators are the right way around: make sure you check the diagram in the data sheet to see if it is a bottom view or a top view!
Yep. When the TO-220 package is laying with the tab down and the writing up, I'm counting left to right starting with 1. (Actually, I ran into datasheets with different numbering conventions -- the reason that things were connected incorrectly the first time around -- and now I'm just thinking Vin, Vout, Vadj which -are- constant across manufacturers on each device.)
-
[
18Vrms AC after full-bridge rectification will give sqrt(2)*18V DC, because the diodes + caps work as peak detectors. The DC voltage is close to the peak AC voltage, which in the case of a sine is sqrt(2) times larger than the RMS voltage that's usually quoted. Subtract about 2V for the drop across the two diodes. You can probably find some illustrated examples on a site like Wikipedia. Note that max current decreases with even more than sqrt(2), because of the uneven loading of the transformer (caps only charge during the peaks). About 0.6 is usually used.
Aah, okay. This makes sense, but I hadn't thought about it.QuoteV^2/R = 2.6W for 24V/220ohm, no wonder that they commit suicide. The 24V output voltage is wrong, however, it should be about 1.25V if R1/R4=0 ohm. Are you sure that the wiring is correct, i.e. no short between the regulators via the heat sink (the case of the LM317 is Vout I believe, but the case of the LM337 is Vin)? Does Vout drop if you decrease R1 at all, i.e. does it work until it goes to zero and then blow up, or is Vout a constant ~24VDC?
Vout does decrease as the pot's decreased, but nowhere in proportion to what it should. At the point that the heat in the sense resistors spikes, I'm getting about 23-24 volts out, and there's very, very little rotation on the pot left at that point. (The resistance at that point across the two terminals of the pot I just tested was about 180 ohms.)
Edit: The pots are linear taper. -
AronCake is full of poorly designed circuits - this is one of the better ones, at least it works. It just needs a bit of tweaking for optimal performance.
Yeah, this one looks pretty much like the reference designs I've seen.QuoteR2 and R3 should be 120R otherwise the output voltage may be too high when unloaded but if you do that you should change the pots to 2.5k.
I originally used 120 ohm resistors but the damned things blew out even faster. QuoteIf you plug the resistor values into the formula given on the LM317 datasheet you get 27.3V but you won't be able to get the voltage as high as this. The maximum output voltage of this circuit when unloaded will be around 24V but there won't be any voltage regulation. The maximum output voltage when fully loaded will be around 15V. This means you'll have a lot of dead band i.e. when it's fully loaded 55% of the pot's adjustment won't make any difference to the output voltage and there'll be no regulation.
I'd recommend increasing the capacitor values to 3300uF, using 1k for the pots and 82R for R2 and R3; this will give a maximum output voltage of 16.5V with minimal deadband at full load. The pots should be rated to at least 250mW.
Thanks! I'll into making those changes once I have a circuit that adjusts properly and doesn't smoke.
-
With the pots cranked all the way up to their maximum resistance, about .635 across R3 and .960 across R2. As the pot is cranked down, the voltage gets up to about 18V and then the resistors start to get really hot.
This is where something is clearly wrong. If the regulators are working correctly the sense resistors should have a constant 1.25 V across them. The regulator is supposed to adjust the output voltage to make this happen.
A simple test would be to work with just the LM317 to start with, connect ADJ directly to the 0 V rail and place 240 ohms between ADJ and Vout. Vout should then be 1.25 V. If it is not, perhaps the regulator is faulty?
Also, are you being sure to place the 1 uF C3 close to the Vin pin of the regulator to make sure it doesn't oscillate?
To help avoid blowing things up, you might disconnect the transformer and replace it with a small 9 V battery while you test voltages and make sure things are working correctly. -
Yep. When the TO-220 package is laying with the tab down and the writing up, I'm counting left to right starting with 1. (Actually, I ran into datasheets with different numbering conventions -- the reason that things were connected incorrectly the first time around -- and now I'm just thinking Vin, Vout, Vadj which -are- constant across manufacturers on each device.)
"tab down" ??
Are you sure you are connecting the regulators the right way round? Just checking because the data sheets usuall show the TO220 devices witht he Tab up.
If you put the TO220 devices down with the tab up and the writing visible (so you are looking at the front, not the heatsink side of the device), then the Adj pin is the first on the left. On the 317, the center pin is the output. On the 337, the right pin is the output.
When it is working, the sense resistors will never have more then 1.2V to 1.3V across them. The moment you see more then that, you know either the regulators are dead or they are connected wrong.
Now once it is working, another concern is the amount of power going into the potentiometers.
If you use a 220 ohm sense resistor with a 5K potentiometer for the LM317, then you will probably be OK.
If you use a 120 ohm sense resistor with a 2.5K potentiometer with the LM337, then it has to be one of the carbon potentiometers that are about 1 inch in diameter. The ones that re about 3/4 inches in diameter or smaller will probably be a bit over their power limits, and it will not be reliable.
In general though, pulling 5 or 10mA from a potentiometer wiper is not a great ideal unless you are using a wirewound potentiometer. Most power supply designs will only be drawing microamps from the potentiometer wiper. The issue is that if the potentiometer starts failing and goes open circuit in places, the output voltage may jump to maximum. (I am not sure if it will jump to maximum or just go to 0V - probably worth a test).
Richard -
If you put the TO220 devices down with the tab up and the writing visible (so you are looking at the front, not the heatsink side of the device), then the Adj pin is the first on the left. On the 317, the center pin is the output. On the 337, the right pin is the output.
Just to mention, but for me 'tab' and 'heatsink' are synonymous. The plastic bit with the writing on is the case. So I guess the important thing is to ignore words and look at the diagrams... -
Quote
R2 and R3 should be 120R otherwise the output voltage may be too high when unloaded but if you do that you should change the pots to 2.5k.
I am pretty sure the data sheet suggests 240 ohm which meets the minimum load current of 5 milliamp.QuoteTo help avoid blowing things up, you might disconnect the transformer and replace it with a small 9 V battery while you test voltages and make sure things are working correctly.
Or put some 1 k / 0.5 watt resistors in series with the rectifier output which should limit the current to something that won't blow anything up. The maximum voltage will be slightly reduced due to the 5 milliamp minimum load current.QuoteNote that the power rating of a pot applies to the pot at max resistance, when all of the resistive material is being used. If you've set it to just 1% of its value, only about 1% of its surface is used, and the power capacity should be derated accordingly.
Luckily for this application the LM317 (when working properly) runs a constant current through the pot. As the resistance is reduced so is the power.QuoteThe issue is that if the potentiometer starts failing and goes open circuit in places, the output voltage may jump to maximum. (I am not sure if it will jump to maximum or just go to 0V - probably worth a test).
It goes to maximum. You should make sure to connect the 3rd unused terminal to the wiper, and put a capacitor to ground which will reduce transients due to wiper chatter.
-
AronCake is full of poorly designed circuits - this is one of the better ones, at least it works. It just needs a bit of tweaking for optimal performance.
I just had a look at the AronCake site for the first time.
You are right - some of those circuits are really bad, but the some of the comments are amazing.
I loved the comment with the LM317/337 supply circuit by someone who wants to build a +/- 110VDc supply. I am not sure if he is going to stick with the LM 317 and 337. His great idea is not to bother with the transformer - just rectify the 220V mains with a bridge rectifier. To get the 0v rail, his will use 2 series filter capacitors on the output of the rectifier, and use the junction of the capacitors for the 0V.
The "Hi Voltage Hi Current" circuit on the site is a microwave oven transformer with a bridge rectifier to give 2KV DC out. At least the editor was kind enough to sayQuoteIf you build this supply without knowledge in electronics or high voltage, you have basically signed your own death certificate.
Thanks
But undaunted, one of the comments is from someone who wants to make his own 100KV to 200KV supply and wants details on how to design the ferrite transformer.
Another person wants the same 2KV supply, but wants to know how to increase it to 5000 to 8000 amps output.
Obviously, these two people need to get together.
What a site!
Richard
-
You've not read the datasheet properly and don't worry, I'm not having a go, most people miss this because the example gives 240R.QuoteR2 and R3 should be 120R otherwise the output voltage may be too high when unloaded but if you do that you should change the pots to 2.5k.
I am pretty sure the data sheet suggests 240 ohm which meets the minimum load current of 5 milliamp.
The application circuit on the datasheet shows the LM117 which has a minimum load current of 5mA. If you look at the specification for the LM317, you'll see it's really 10mA.
http://www.national.com/ds/LM/LM117.pdf
The LM337 datasheet actually specifies 120R.
http://www.national.com/ds/LM/LM137.pdf
-
Huh. I never noticed that
The typical value is 3.5 but the worst case value is 10 mA, while the plot shows the typical variation from 1 to 4 mA based on voltage drop and temperature. I've made lots of LM317 based regulators with 240 ohm and never had a problem, but I guess I will change my ways
-
Huh. I never noticed that
It's the LM337 that needs the higher load current, not the LM317! The typical value is 3.5 but the worst case value is 10 mA, while the plot shows the typical variation from 1 to 4 mA based on voltage drop and temperature. I've made lots of LM317 based regulators with 240 ohm and never had a problem, but I guess I will change my ways -
It's the LM337 that needs the higher load current, not the LM317!
I suggest you read the datasheet too, especially the extract attached to my previous post.
The LM317 and LM337 both have the same minimum load current.