regulated supply oscillation or whatever that is - Page 2 - diyAudio
Go Back   Home > Forums > Amplifiers > Power Supplies

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 14th March 2014, 07:51 PM   #11
Trileru is offline Trileru  Romania
diyAudio Member
 
Join Date: Dec 2010
Quote:
Originally Posted by sghr220 View Post
I suggest (but don't take me too seriously ) removing the output snubber and R.load and doing this test again.

Best regards.
Actually I've made the test before connecting the snubber. It made no difference, looked the same before/after
  Reply With Quote
Old 14th March 2014, 10:43 PM   #12
Trileru is offline Trileru  Romania
diyAudio Member
 
Join Date: Dec 2010
Quote:
Originally Posted by funk1980 View Post
A board layout would be very helpful.
Attached Images
File Type: png rg.png (80.3 KB, 81 views)

Last edited by Trileru; 14th March 2014 at 10:46 PM.
  Reply With Quote
Old 14th March 2014, 10:48 PM   #13
Trileru is offline Trileru  Romania
diyAudio Member
 
Join Date: Dec 2010
Quote:
Originally Posted by esgigt View Post
Trileru, from your description I get the impression that beyond 1A the stability margins of the regulator are exceeded.
You can solve this in several ways:


Or you can apply a small series resistor ( .2 a .3 Ohms) and some 10 to 20 mF on the output line. This will give less load regulation, but better ripple and noise performance.
I just mentioned that there is a 0.47R resistor and 47uF capacitor in series on the output. It made no difference with/without.

After I make the test with more resistance on the output I will post the result.
  Reply With Quote
Old 15th March 2014, 12:07 AM   #14
diyAudio Member
 
funk1980's Avatar
 
Join Date: Sep 2011
Location: Groningen
Just some thoughts. Might this just be the normal operating characteristics of the 338? The frequency is a doubling of the signal used for testing. Equal amounts of current is drawn for both the up-going and down-going part of the sinewave. That's the doubling. Edit: You can check this by measuring the input signal too and check if the vallies of the PSU's output correspond with the peak-peaks from the sinewave.

The LM338 itself has a typical load regulation of 0.3% over the full temperature range (1% max). 48mV is 0.29% of the 16.5V output. That's within spec. Heat is a factor here. Your heatsink seems rather small. I don't know the input voltage, but let's assume it's a 7.5K/W type heatsink + the TO220's thermal resistance (1K/W) + the goop/silicone, that's another 0.5K/W, making a total of 9K/W. If let's say the LM338 has to drop 5 volts at 2 amps, that's 10W. Raising the temperature 90K above ambient, usually 25C, making 115C!! So the LM338 becomes hot! And line regulation gets worse.

Concerning the PCB layout, there's a lot of possibility for ground loops and high current paths going past sensitive circuitry. The effects become increasingly worse when more current is drawn. Ground planes are not an easy one-for-all solution. Currents flow through the ground-plane and sometimes through (sensitive) parts where you don't want it to. I learned that the hard way not long ago. Be mindful with component placing. For example, the diodes are not critical for normal operation, so their spot would be better suited for the bias network.

So things to improve and perhaps increase the line regulation? You could increase the output capacitance, making shure the stability is maintained. Better cooling? Try force-cooling the heatsink (fan) and see if it improves. Maybe shuffle the board layout a bit, keeping ground currents in mind. But for the most part, I believe this is just the rail showing a normal changing load.

Oh and why the 270ohm load resistor? The LM338 minimum load is 3.5mA (10mA max), which is easily drawn by the bias network alone.

Last edited by funk1980; 15th March 2014 at 12:19 AM.
  Reply With Quote
Old 15th March 2014, 12:42 AM   #15
Trileru is offline Trileru  Romania
diyAudio Member
 
Join Date: Dec 2010
Thank you for the info.
I'm actually using a LT1084 with the TDA amp. And my heatsink has a 4.5K/W rating. Rectified voltage is 22.5VDC, so -1.5VDC for the regulator -16.5VDC output leaves 4.5VDC on the regulator to be burned. At 2A that's 9W. At 1A it's only half of that, so with 6K/W rating that's only 27 degrees + 25 room temp. So 52 or even 55 degrees C is not much. At full 2A it would theoretically go up to 79-80 degrees. No issues with heating. I'm more concerned with the rectifying diodes as they drop 4V at 2A, so that's 2W per diode. I can't test for more than a few seconds as I don't have a heatsink for them and I won't be bothered to make one as they will probably never need it. I don't think that they will ever see 1W continuously each. But my layout permits such a heatsink for those diodes if need be anyway
I also separated the ground planes to keep the currents from going crazy. There's a split in the middle with only a thin portion to tie the rectifying part from the regulated part. And also I separated C5 from the ground plane, only to join at the output.
For the minimum current I was aiming for about 50mA just to keep the regulator happy (lt108x is around 5-10mA minimum current draw). And I didn't have any 330ohm resistor in a larger package so I used what I had at hand. No worries, 60mA is ok. That's about 1W and I put a 5W resistor.
LT1084 line regulation is about 0.035% for Vin-Vout = 5V, and I have about 6V. And line regulation is guaranteed for the whole power dissipation of the device, and for LT1084 for Vin-Vout=5V the current limit is of 6.5A, or about 30W.
quote from the datasheet:
Note 3: Line and load regulation are guaranteed up to the maximum power
dissipation (60W for the LT1083, 45W for the LT1084 (K, P), 30W for the
LT1084 (T) and 30W for the LT1085). Power dissipation is determined by
the input/output differential and the output current. Guaranteed maximum
power dissipation will not be available over the full input/output voltage
range.
So 48mV or 0.29% of output is an order of magnitude worse. Although it is stated that worst case scenario is 0.2% but typically 0.035-0.05%.
And I see no change between cold start and later on while testing.
I made the measurements with ground on C5 negative leg and signal on C6 positive leg.

P.S.
This is the heatsink that I'm using:
http://www.tme.eu/en/details/sk129-6...r-elektronik/#
  Reply With Quote
Old 15th March 2014, 01:20 AM   #16
esgigt is offline esgigt  Netherlands
diyAudio Member
 
Join Date: Dec 2013
Quote:
Originally Posted by Trileru View Post
I just mentioned that there is a 0.47R resistor and 47uF capacitor in series on the output. It made no difference with/without.

After I make the test with more resistance on the output I will post the result.
That is not how I interpreted the schematic you provided. The way I interpreted your schematic is that the resistor/capacitor (.47R/47uF) is in parallel to the load.

What I tried to suggest is a series resistor (.2/3 Ohms) following the regulator with a large capacitor shunting the load. I did not mean the .47R/47uF snubber
  Reply With Quote
Old 15th March 2014, 01:34 AM   #17
diyAudio Member
 
funk1980's Avatar
 
Join Date: Sep 2011
Location: Groningen
Ah ok, I missed the part for LT1084 vs. LM338 . However, we're talking load regulation (made the typo myself just now). Load regulation for the 1084 is between 0.2% and 0.4% @ 25C. The 0.015 is line regulation; how well it withstands changes in input voltage. So your 0.29% is well within spec.

Concerning the diodes. They don't have a sinusoidal current through them, but the loading peaks produced by the caps instead. Most of the time, the datasheet has a separate, usually higher, current rating for this.

The temperature resistance given by LT is a very realistic. Worst case scenario is 2.7K/W. I'm not saying the 1084 is getting too hot, just that it all adds up. At 1 amp, we could potentially see a junction temp of 60C. If the PSU is put inside a small enclosure, ambient will rise to etc. etc. Again, load regulation gets worse with temperature.

Check the current path of the adjust pin cap and bias network for example. They're tied to the common output pin. That's a high current point!
  Reply With Quote
Old 15th March 2014, 01:37 AM   #18
infinia is offline infinia  United States
diyAudio Member
 
infinia's Avatar
 
Join Date: May 2005
Location: Californication
can you show the PCB reverse side as well. There should be a split on the ground side mirroring the top side. otherwise two overlapping ground planes form a pretty good coupling capacitor, contaminating the output ground with fast charging pulses.
big symptom > doubling on the AC signal points to diodes. Test with a big enough static output and load look for 100 Hz stuff on the outputs. Remember common mode ripple and noise is untouched by the regulator. My experience has been using ground planes are nothing but trouble for DC power and/ or audio power amps. DC headroom looks OK I guess, measurements preferred over load analysis. how about AC waveforms with a dummy load applied?
__________________
like four million tons of hydrogen exploding on the sun
like the whisper of the termites building castles in the dust
  Reply With Quote
Old 15th March 2014, 01:02 PM   #19
Trileru is offline Trileru  Romania
diyAudio Member
 
Join Date: Dec 2010
Quote:
Originally Posted by infinia View Post
can you show the PCB reverse side as well. There should be a split on the ground side mirroring the top side. otherwise two overlapping ground planes form a pretty good coupling capacitor, contaminating the output ground with fast charging pulses.
Red is top side and is ground, and bottom is blue and is power plane and stops at the regulator. I don't know how to show the bottom side as all the information is already in that picture Later on, after I replace the oil in my bike and give it a nice wash I will add some more capacitance on the output of the board and make the tests. I have a couple of 8ohm 100W resistors.
  Reply With Quote
Old 15th March 2014, 07:16 PM   #20
diyAudio Member
 
Join Date: Dec 2012
Location: Buenos Aires
Quote:
Originally Posted by Trileru View Post
Red is top side and is ground, and bottom is blue and is power plane and stops at the regulator. I don't know how to show the bottom side as all the information is already in that picture Later on, after I replace the oil in my bike and give it a nice wash I will add some more capacitance on the output of the board and make the tests. I have a couple of 8ohm 100W resistors.

Hi, as I stated earler in the thread I think you will benefit from extra capacitance (large value) on the output side if you install a high current and gain device like a darlington transistor. That way you will unload the regulator transfering the job to the pass device. It will also take care of the large capacitance charging current and keep the voltage value steady.
__________________
Pablo
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Regulated Vs Non-regulated linear Power Supply mbeards Tubes / Valves 2 30th January 2013 10:45 PM
Power supply oscillation and Class D? eem2am Class D 2 27th August 2011 09:37 PM
3 Pin regulated supply VS zener based supply KevinLee Solid State 3 15th September 2008 03:40 PM
Regulated power supply meanman1964 Everything Else 13 9th April 2007 03:50 AM
Supply depended oscillation Thoru Solid State 15 16th February 2007 11:49 AM


New To Site? Need Help?

All times are GMT. The time now is 10:38 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2