| garbage |
hi all
i've built 2 regulated psu using pedga's schematic.
only difference is that i am using +12/-12 vac transformer secondaries to get a regulated +15/-15vdc. also, in place of the 2.2k resistor, i had a variable resistor. higher capacitance was also used. a diode was also placed in the direction of ADJ to OUT pin of lm338t.
one of them works fine, the other(built at a later date), does not seem to work correctly.
the voltage will fluctuate from 3+v to 8+v then drop to 3+v then back again. between the cycle, it will hold steady for a short while at 3+v and also at 8+v.
after rectifier stage is ok, measured close to 19vdc. somewhere from lm338t onwards, there seem to be a problem. i've checked my connections. they seem fine.
any ideas where i might have possibly gone wrong? |
|
|
| li_gangyi |
| could it be a bad potentiometer?? perhaps u're cycling it into some kinda protection...and what output volatges are u aiming for? |
|
|
| janneman |
What voltage do you measure at the adj pin for the one that has the problem? If that fluctuates, try it with the adj cap disconnected, it can be leaky (or reversed). Also, check/remove temporarily any diodes at the adj pin to make sure they are not leaking or something. Does the reg get warm at all?
Jan Didden |
|
|
| garbage |
i've taken out the variable resistors out for measurment, they seemed ok. i'm actually aiming for +/-15vdc.
the first circuit that i built worked, it is now powering my opa627/buf634. since i have 2 transformers of the same type, i wanted to build another one so that i can use one for each channel.
 |
|
|
| garbage |
| quote: | Originally posted by janneman
What voltage do you measure at the adj pin for the one that has the problem? If that fluctuates, try it with the adj cap disconnected, it can be leaky (or reversed). Also, check/remove temporarily any diodes at the adj pin to make sure they are not leaking or something. Does the reg get warm at all?
Jan Didden |
i'll try out your suggestions.
thanks. :)
edit:
the regs do not get warm, and i've tried with the adj cap disconnected(no difference).
next to try is to remove adj cap and diode. |
|
|
| janneman |
What's the voltage on the adj pin doing. Try to measure the voltage between the output pin and the adj pin of the reg. If that remains constant about 1.25V, than the reg is OK.
To measure is to know!
Did you change the divider on the rteg output, because for the 15V version you need to change it from what you showed in the schematic upstairs.
Jan Didden |
|
|
| carlosfm |
Check the pinout of the regs.
Look at it and check if you made the right connections to the right pins. |
|
|
| joensd |
With 2*12VAC youīll get about +-18VDC maximum depending on your transformer.
Aiming for +-15VDC isnīt a problem in most cases (Iīm using the same setup with 7815,7915 and it works nicely) but the voltage-headroom is very small and regulation might get worse or donīt work properly. Donīīt know the LM338 too well though.
If you donīt need so much voltage swing set the resistors up for +-12VDC to see if itīs that.
But check the more obvious things first. |
|
|
| jackinnj |
| just a silly question and I cringe asking it, but did you insulate the LM338's from the heat sinks? |
|
|
| drdagor |
Your system is probably oscillating. Because the capacitors are so large it doesn't really look like an oscillation, but that's probably what you have. With 4700 uf and almost no load you probably have a time constant of minutes.
Bypass capacitors is the normal fix for this.
In addition to the capacitors you have, it would be worthwhile to add a 10uf and a 0.1uf bypass capacitor to the regulator mounted as close to the regs as possible. The 10uf should be tantalum--the higher the voltage the better (since at the same uf value, the higher voltage ones have lower ESR).
Normally the 0.1 goes on the input side of the reg and the 10uf on the output side. But in your case, because the input capacitance is so huge on the input side, you may want to put a 10uf on both the input and output side.
Hope this helps. |
|
|
| carlosfm |
You need to!:bawling: :bawling: :bawling:
Good one, man...:D |
|
|
| carlosfm |
My post above was just after this one when I posted:
| quote: | Originally posted by jackinnj
just a silly question and I cringe asking it, but did you insulate the LM338's from the heat sinks? |
Strange... now it isn't.:eek: :scratch1: |
|
|
| garbage |
hi guys, thanks very much for all your input.
| quote: | Originally posted by janneman
What's the voltage on the adj pin doing. Try to measure the voltage between the output pin and the adj pin of the reg. If that remains constant about 1.25V, than the reg is OK.
To measure is to know!
Did you change the divider on the rteg output, because for the 15V version you need to change it from what you showed in the schematic upstairs.
|
i'll check the voltage on the adj to out pin tonight.
i did not change the resistors as they gave me the output voltage that i need on the psu that is working.
| quote: | Originally posted by carlosfm
Check the pinout of the regs.
Look at it and check if you made the right connections to the right pins. |
i'll re-check that again. but i am pretty sure that it is right. from the image above, it is like this (from top up)
+19v o input
o output
o adj
| quote: | Originally posted by jackinnj
just a silly question and I cringe asking it, but did you insulate the LM338's from the heat sinks? |
they are insulated with those thermal pads.
the heatsink also do not show continuity when i used a multi-meter to check them.
| quote: | Originally posted by drdagor
Your system is probably oscillating. Because the capacitors are so large it doesn't really look like an oscillation, but that's probably what you have. With 4700 uf and almost no load you probably have a time constant of minutes.
Bypass capacitors is the normal fix for this.
|
i have one 0.1uf cap soldered after the rectifier and one 0.1uf soldered as the last cap on the output. these are under the board. |
|
|
| jackinnj |
| quote: | Originally posted by drdagor
Your system is probably oscillating. Because the capacitors are so large it doesn't really look like an oscillation, but that's probably what you have. With 4700 uf and almost no load you probably have a time constant of minutes. |
that doesn't make any sense at all, none, zero, zip, nada... |
|
|
| jackinnj |
| quote: | Originally posted by garbage
h
they are insulated with those thermal pads.
the heatsink also do not show continuity when i used a multi-meter to check them. |
ok, you used an insulated pad...
... but did you use an insulated "shoulder washer" -- if you just screwed them through, even with the insulated pads there may be more than intermittent contact between the output tab and the heat sink. |
|
|
| garbage |
| quote: | Originally posted by jackinnj
ok, you used an insulated pad...
... but did you use an insulated "shoulder washer" -- if you just screwed them through, even with the insulated pads there may be more than intermittent contact between the output tab and the heat sink. |
actually i realised that when i was attaching the heatsinks.
as the heatsink shows no continuity, i thought it'd be fine.
i'll remove the heatsinks and test them tonight just to be sure.
thanks. |
|
|
| carlosfm |
You need to use plastic washers on the screws.
You have the screws in contact with the chips and with the heatsink.
Even with those isolators, always check for continuity between the chip and the screw. |
|
|
| garbage |
took out the heatsinks last night and tried the supply again.
only getting about 3.5v or so. :whazzat:
voltage across adj and output pin was in the millivolts range.
i'm gonna remove everything from the lm338t onwards over the weekend and solder the components one by one, testing at each stage to see if it works. :smash: |
|
|
| sek |
| quote: | | voltage across adj and output pin was in the millivolts range. |
The regulator should keep a constant reference voltage of around 1.25V between OUT and ADJ.
A true voltage between ADJ and OUT in the millivolt range might sound like a damaged chip.
How did the soldering work? Did you probably overheat the LM338? Or overload it?
| quote: | | i'm gonna remove everything from the lm338t onwards |
You shouldn't leave the chip unconnected. At least use a replaced voltage divider network, no ADJ bypass cap and no OUT bypass cap for a short measurement. With an appropriate input voltage of >15V, the regulator should really show some sensible behaviour now. Otherwise it might be damaged...
| quote: | | i did not change the resistors as they gave me the output voltage that i need on the psu that is working. |
Just to make shure I understand you right: You use a network of 120R and 2.2k according to Pedja's circuit?
Check with the datasheet: On page 4 it shows the equation to calculate the resistor network for a required output voltage. Using 120R as the upper and 2k2 as the lower resistor would calculate to Pedja's required value of around 25V.
For 15V and 120R as the upper resistor, the equation leads to a value for the lower resistor of ca. 1.3k ... :smash:
If you really used the wrong values here, that would explain your problem. It would also explain the too low reference voltage between ADJ and OUT, because for Vin < (Vout+Vdrop) the regulator is fighting against itself, showing ill behaviour at the OUT pin. That makes me wonder why the other board works... :xeye:
Sebastian. ;) |
|
|
| carlosfm |
| quote: | Originally posted by sek
How did the soldering work? Did you probably overheat the LM338? Or overload it?
|
They were just not isolated from each other.
As this PSU uses two positive regulators to make a +, G and -, you can imagine what went wrong!:hot: :hot: :hot: |
|
|
| sek |
| quote: | | you can imagine what went wrong! |
Yeah, although it hurts. But why should the other one work, then.
Did he insulate the ICs and get the voltage divider correct on the first board, just to make both wrong on the second? I couldn't imagine...
Good Luck, Garbage. ;)
Just don't bother with that LM338. Replace it with a new one and check the old one later. It might still work, but you don't want to determine that the hard way. The part is too cheap to spend more than an evening just to rescue a probably damaged one.
Sebastian. |
|
|
| garbage |
| quote: | Originally posted by sek
You shouldn't leave the chip unconnected. At least use a replaced voltage divider network, no ADJ bypass cap and no OUT bypass cap for a short measurement. With an appropriate input voltage of >15V, the regulator should really show some sensible behaviour now. Otherwise it might be damaged...
|
ok, will try this.
| quote: | Originally posted by sek
Just to make shure I understand you right: You use a network of 120R and 2.2k according to Pedja's circuit?
Check with the datasheet: On page 4 it shows the equation to calculate the resistor network for a required output voltage. Using 120R as the upper and 2k2 as the lower resistor would calculate to Pedja's required value of around 25V.
For 15V and 120R as the upper resistor, the equation leads to a value for the lower resistor of ca. 1.3k ... :smash:
If you really used the wrong values here, that would explain your problem. It would also explain the too low reference voltage between ADJ and OUT, because for Vin < (Vout+Vdrop) the regulator is fighting against itself, showing ill behaviour at the OUT pin. That makes me wonder why the other board works... :xeye:
Sebastian. ;) |
hmm, but i had a 2k2 variable resistor and i tried turning them to one extreme... perhaps i can try the other extreme.
the other psu worked by having output of some low voltage up to about 18 or 19vdc, where further turns on the variable resistor will not yield any increase as that is the max after the rectifiers.
thanks.
| quote: | Originally posted by carlosfm
They were just not isolated from each other.
As this PSU uses two positive regulators to make a +, G and -, you can imagine what went wrong!:hot: :hot: :hot: |
i did not terminate the ends of the output to form a common ground.
so the circuits are effectively 2 +15vdc output. just that i reversed the other one as -15vdc for my buffer. |
|
|
| li_gangyi |
| no...the isolation meant that u didn;t seperate the 2 ICs from the same heatsink...they are conducting...forming a short somewhere...or so he presumes... |
|
|
| garbage |
oic.
actually the 2 heatsinks are covered with thermal pad on one side. and they(heatsinks) do not conduct electricity.
each ic had it's own heatsink. |
|
|
| li_gangyi |
| a pad alone is not enough...since the screw is there to conduct...do u use insulating washers?? |
|
|
| garbage |
nope.
but even removing the heatsink does not help.
i'll see how it goes during the weekends when i have time to play around with it. :smash: |
|
|
| garbage |
hi all
played around with my psu, and found out that the R1 resistors that i bought in a hurry are not 120R but 120k!!
also one of the psu's variable resistor's pin 2 to 3 is not working, so i had to use pin 1 to 2 instead.
here is one section of the psu running at 15vdc output. (notice the orientation of the variable resistor is switched as one of them is using pin 1 and 2 instead)
and here are the 2 psu powering my line drive (opa627/buf634)
played john neptune's asian roots cd, track 3 - jegoging on it.
separation and depth are simply great. i did not want to switch to another cd. got to go and listen more.
cya, and thanks for all the suggestions. :) |
|
|
| sek |
That's great news! :D
Good to hear that You didn't accidentally wreck nearly every second component in it. ;)
But now that You show it off: what's that preamp? |
|
|
| garbage |
hahaha... yup, i must be lucky it's only the psu section.
the preamp is originally Line Drive by The Mod Squad. it is a passive pre. now Mod Squad is now known as McCormack. they no longer produce passive pres.
i have added an opamp/buffer into it and it now has dual external psu for each channel. i will be getting a case for the external supplies soon. |
|
|
|
