Hi Imix,
sorry, I never noticed the Amps output column.
I thought you were posting a constant 4k7 load and varying the input voltage to show output dependence on input voltage.
Now that I read your results properly you have reasonable regulation and still have a margin of 6.6volts across the regulator for Vdrop +half the ripple.
It would be nice to know what the ripple is at 5.1A output. I suspect it is pretty high.
It also looks like 24V output is close to the maximum you can get with this transformer at full output current. You will be able to get a much higher ripple free output voltage at lower currents but you run the risk of putting hum into your load without noticing.
sorry, I never noticed the Amps output column.
I thought you were posting a constant 4k7 load and varying the input voltage to show output dependence on input voltage.
Now that I read your results properly you have reasonable regulation and still have a margin of 6.6volts across the regulator for Vdrop +half the ripple.
It would be nice to know what the ripple is at 5.1A output. I suspect it is pretty high.
It also looks like 24V output is close to the maximum you can get with this transformer at full output current. You will be able to get a much higher ripple free output voltage at lower currents but you run the risk of putting hum into your load without noticing.
So the board handles 5A ok, eh? 
I can take my board in to work and measure the parasitic capacitance between the input, output, and ground connections just to see what they are. Just to note, since the board has a ground plane on the bottom side and input and output planes on the top side, the parasitic capacitances between the planes to ground will just add to larger caps and bypass caps that are on the board - meaning that they shouldn't cause the regulator to oscillate.
What kind of heatsink are you using for the 338 and is it electrically connected to anything? The output tab of the regulator is at the output voltage so you can't use a grounded heatsink. Even using a silpad to isolate the regulator from a grounded heatsink could cause the regulator oscillate. I've seen that happen before!
I can take my board in to work and measure the parasitic capacitance between the input, output, and ground connections just to see what they are. Just to note, since the board has a ground plane on the bottom side and input and output planes on the top side, the parasitic capacitances between the planes to ground will just add to larger caps and bypass caps that are on the board - meaning that they shouldn't cause the regulator to oscillate.What kind of heatsink are you using for the 338 and is it electrically connected to anything? The output tab of the regulator is at the output voltage so you can't use a grounded heatsink. Even using a silpad to isolate the regulator from a grounded heatsink could cause the regulator oscillate. I've seen that happen before!
Here are some numbers...
With the measuring probes at the input and output through holes (where the wires are soldered) there is: 22.68pF between Vin and GND, 23.73pF between Vout and GND, and 11.24pF between Vin and Vout.
With the measuring probes at the regulator through holes (where the regulator pins are soldered) there is: 20.53pF between Vin and GND, 21.55pF between Vout and GND, and 10.57pF between Vin and Vout.
Clearly, a minimal amount of stray capacitance that will be more than swamped out by a bypass capacitor let alone a filter cap.
With the measuring probes at the input and output through holes (where the wires are soldered) there is: 22.68pF between Vin and GND, 23.73pF between Vout and GND, and 11.24pF between Vin and Vout.
With the measuring probes at the regulator through holes (where the regulator pins are soldered) there is: 20.53pF between Vin and GND, 21.55pF between Vout and GND, and 10.57pF between Vin and Vout.
Clearly, a minimal amount of stray capacitance that will be more than swamped out by a bypass capacitor let alone a filter cap.
Hey guys,
AndrewT, sorry about the confusion. Is there a god way to measure ripple without a scope? If ripple appears it will be too high at high voltages/currents, what is the best way to fix it? Larger filter caps? A choke maybe?
BWRX, hey man, the board seems to handle 5A just fine- although I probably wouldn't run it all day at 5A. The board was never on my list of culprits, layout is great on it. I'm just wondering if having 4 inch leads going to the reg from the board is setting up an oscillation.
I am using the TO3 version on the reg. It is mounted on a 6 x 3 aluminum heatsink with fins on both sides orriginally used for a 100A stud rectifier and set off the enclosure by nylon hardware. I checked for continuity on the highest range with my fluke 87 (before I blew it up on another project) and it was super high between all pins and the heatsink/case. The device is insulted by mica and thermally connected with some goop.
Thanks!
AndrewT, sorry about the confusion. Is there a god way to measure ripple without a scope? If ripple appears it will be too high at high voltages/currents, what is the best way to fix it? Larger filter caps? A choke maybe?
BWRX, hey man, the board seems to handle 5A just fine- although I probably wouldn't run it all day at 5A. The board was never on my list of culprits, layout is great on it. I'm just wondering if having 4 inch leads going to the reg from the board is setting up an oscillation.
I am using the TO3 version on the reg. It is mounted on a 6 x 3 aluminum heatsink with fins on both sides orriginally used for a 100A stud rectifier and set off the enclosure by nylon hardware. I checked for continuity on the highest range with my fluke 87 (before I blew it up on another project) and it was super high between all pins and the heatsink/case. The device is insulted by mica and thermally connected with some goop.
Thanks!
Hi Imix,
don't apologise, you did nothing wrong. You are allowed to ask questions, that's the main purpose of this forum - to enlighten.
More to the point, do you understand why the reg appeared to be inoperative when you asked for output voltages above the 24V you now have?
Another point worth mentioning. What transformer voltage are you using. I guess it's 24Vac (when loaded) or about 25Vac when unloaded.
This is a general rule for regulated supplies.
AC input voltage ~= regulated DC output voltage for supplies in the range 12V to 30V.
At lower voltage the AC needs to be a bit higher and at higher voltage you can work with slightly lower AC voltage input.
don't apologise, you did nothing wrong. You are allowed to ask questions, that's the main purpose of this forum - to enlighten.
More to the point, do you understand why the reg appeared to be inoperative when you asked for output voltages above the 24V you now have?
Another point worth mentioning. What transformer voltage are you using. I guess it's 24Vac (when loaded) or about 25Vac when unloaded.
This is a general rule for regulated supplies.
AC input voltage ~= regulated DC output voltage for supplies in the range 12V to 30V.
At lower voltage the AC needs to be a bit higher and at higher voltage you can work with slightly lower AC voltage input.
AndrewT, that is for certain on this forum- enlightening.
I'm pretty sure I understand what is going on, too much voltage sag and too high ripple is causing the reg to reach an output ceiling before it should.
The transformer is 27v unloaded, 26.5v with a 1amp load.
Oh, and it's rated at 100va, I realize this is a little small for a 5A psu.
So, what's the solution besides maybe a little larger transformer?
Is 18K uf not enough for 5A at 30V?
Thanks.
I'm pretty sure I understand what is going on, too much voltage sag and too high ripple is causing the reg to reach an output ceiling before it should.
The transformer is 27v unloaded, 26.5v with a 1amp load.
Oh, and it's rated at 100va, I realize this is a little small for a 5A psu.
So, what's the solution besides maybe a little larger transformer?
Is 18K uf not enough for 5A at 30V?
Thanks.
Hi,
when a transformer is loaded with a rectifier and smoothing capacitor, it's rating drops to about 70%.
You were asking for about 30V and 4A from the output =120Watts.
Before the reg, you need about 35v @ 4A =140W, now divide by 0.7 to find your minimum VA rating >=200VA.
If you want 5A increase by 25% and for 6A by +50%.
If you decide to use higher than 25Vac then 2 times Vac times Iout ~= VA rating
when a transformer is loaded with a rectifier and smoothing capacitor, it's rating drops to about 70%.
You were asking for about 30V and 4A from the output =120Watts.
Before the reg, you need about 35v @ 4A =140W, now divide by 0.7 to find your minimum VA rating >=200VA.
If you want 5A increase by 25% and for 6A by +50%.
If you decide to use higher than 25Vac then 2 times Vac times Iout ~= VA rating
imix500 said:
Damn? Are you really connecting the regulator to the PCB with a 4 inches long (10 centimeters!!) lead for each pin and expecting any good performance from it?
You should read the application hints on the datasheet very carefully. The most probable causes of your problem are both oscillation and the voltage drop across the output lead being substracted from the 1.25V reference, so that such voltage drop will appear on the output multiplied by the reference resistor divider ratio. You need at least 4 wires (two of them for the output pin, one conneced to each screw of the TO-3 can) and you should keep them definitely shorter than 10cm.
This picture shows my wiring, don't blame me because I did it when I was only 17 years old (and it has gathered a lot of dust since then):
An externally hosted image should be here but it was not working when we last tested it.
Red wire is input. Brown wire is output and is connected to the lower screw. Green wire is output sensing and is attached to the upper screw. Black wire is reference.
Like Eva said, the way you have the TO3 package wired up may be part of the reason the regulator is acting the way it is. The board is made for TO220 package regulators, and there are a bunch out there than can do 5A or more. The LM338 being one of them. I would get the LM338 in the TO220 package along with a larger transformer if you want the maximum current capability without voltage sag.
AndrewT, seems I'll be shopping for surplus transformers tonight! By the way, if I read it correctly the data sheet specs say that current limiting can kick in if there is a difference of 10 volts or less between input and output. Does this sound right?
Eva, the leads can certainly be shortened to about 1 inch I'd guess. I remember well reading about the need for two leads from the output as to compensate for long leads from the device as well as connecting the current set resistor between the pins of the device, but I didn't think 4 inches would be enough to impact it. Guess I was wrong.
So, I'll get a larger transformer ordered and in the mean time shorten the leads as much as possible and add a lead from the output.
Thanks!
Eva, the leads can certainly be shortened to about 1 inch I'd guess. I remember well reading about the need for two leads from the output as to compensate for long leads from the device as well as connecting the current set resistor between the pins of the device, but I didn't think 4 inches would be enough to impact it. Guess I was wrong.
So, I'll get a larger transformer ordered and in the mean time shorten the leads as much as possible and add a lead from the output.
Thanks!
It certainly wouldn't be a bad idea to try the TO220 variant as the board is obviously optimized for that package. That way you could either bolt it to the heatsink vertically or bend the leads so the regulator tab is parallel to the board and mount it to a heatsink that way.
For now, shortening the leads attached to the regulator as much as possible is the best thing you can do to improve performance.
Just yesterday I built up two regulator circuits using LT1085 regulators. I havent tested the current capability yet (the LT1085 is a 3A regulator), but the voltage output is very stable into an 8ohm load at 12V output even down to 1.5V dropout!
I've also got a pair of LT1083 regulators (7.5A continuous!!!) that I thought about using to make a beefy bench supply. They're in the TO-3P package though (looks like a TO220 on steroids ).
If you've got some extra cash to spend I highly recommend the Linear Tech devices. Although the National devices work well too and are cheaper.
For now, shortening the leads attached to the regulator as much as possible is the best thing you can do to improve performance.
Just yesterday I built up two regulator circuits using LT1085 regulators. I havent tested the current capability yet (the LT1085 is a 3A regulator), but the voltage output is very stable into an 8ohm load at 12V output even down to 1.5V dropout!
I've also got a pair of LT1083 regulators (7.5A continuous!!!) that I thought about using to make a beefy bench supply. They're in the TO-3P package though (looks like a TO220 on steroids
).If you've got some extra cash to spend I highly recommend the Linear Tech devices. Although the National devices work well too and are cheaper.
Attachments
These LM338K ICs should provide very good regulation when properly used. Mine were providing +-41.5V for 4 TDA7294 chip-amps, which were driving two 8 ohm and two 4 ohm speakers. I was unable to make the regulators fold back with music, and output voltage drop at full volume mas low enough to be hard to measure with a standard multimeter.
Hi!
While waiting for your new parts, if you feel like experimenting:
Put another set of 0.1uF capacitors on the input, output and Vadj pins to ground, and situate them right where the regulator is. In other words, solder directly onto the TO-3's pins. That should kill any oscillation problems. You should also use a thicker braided wire for the ground reference to the TO-3.
Cheers!
While waiting for your new parts, if you feel like experimenting:
Put another set of 0.1uF capacitors on the input, output and Vadj pins to ground, and situate them right where the regulator is. In other words, solder directly onto the TO-3's pins. That should kill any oscillation problems. You should also use a thicker braided wire for the ground reference to the TO-3.
Cheers!
BWRX, I might add a few to220 versions to the order. They are certainly a LOT less expensive than the to3 package. I have not used any regs other than the LM* products. I'm now curious on what else is out there.
Eva, I've had good results with nationals regs as well. The lm338 I haven't used in over 10 years. I'm revisiting this supply I built when I was 16-17. I kept blowing the current reference resistor back then so I abandoned it.
clem_o, I will try that soon. On the grounding strap, since the ground reference is through a pot parallel with a 22uf cap, would it make a difference?
Thanks!
Eva, I've had good results with nationals regs as well. The lm338 I haven't used in over 10 years. I'm revisiting this supply I built when I was 16-17. I kept blowing the current reference resistor back then so I abandoned it.
clem_o, I will try that soon. On the grounding strap, since the ground reference is through a pot parallel with a 22uf cap, would it make a difference?
Thanks!
Hello all, so it took me a little while to find a transformer. It's 20-0-20 500va. After rectification I'm looking at about 57.6v. This is a bit high for the lm338 right? What's the best way to drop some voltage without taking the transformer apart? I've been thinking about a string of diodes. Any thoughts? Thanks.
Hi, I'm guessing I would not using the center tap- the voltage would be too low. However I am willing to hook it up a variety of ways if it means a lower voltage. If I'm reading the sheet right it looks like the maximum voltage difference between in and out is 42v. So with the 1.2v minimum output thats 43.2v.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.