Modern 317/337 alternatives? - Page 6 - 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 27th February 2013, 12:46 AM   #51
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
Quote:
Originally Posted by maxw View Post
Looks good! Are you actually going to build it?
I think so! I've just revised the layout to add the 317/337 and a couple of other tweaks. I'll probably send the board out to fab in a week or so whenever I figure out what to add in the remaining board space.

I'm kind of liking the 317/337 pre-regulator thing, the more I think about it today. It guarantees that the input of the LT1963A will never exceed around 17.5Vdc + 0.3Vdc. I was concerned that the diode fed cap multiplier might exceed that with power line fluctuations. Also a bit worse than it seems since from the data sheet fine print notes that 20Vdc maximum input is measured from the output. So if the other rail drops out and the schottky clamps it at -0.2Vdc, that would be 19.8Vdc maximum in. With silicon clamp diodes on the rails it would drop to 19.3Vdc or so. Plus I like to add at least a volt of margin there for part tolerances.

The arrangement uses each part for what it does best. The floating 317/337's float while the ground-referenced LT parts pin down the endpoint voltage. I forgot to mention that the indicator LED resistors on each rail are reduced from 4.7K to 2.7K to pull 5mA through the LEDs. The LM parts need around 5mA minimum load with the 340mA flow but the LT parts only consume about 1mA quiescent. So the LEDs are *required* as load for the LM parts, to take care of the case of the output unhooked to anything.

I've also added 0.5R more in series with the 100uH coils to lower the Q of the CLC filter to nuke some peaking in the sims below. With just the coils and their inherent 0.2R the response peaks slightly. I prefer things on the over-damped side.

Should be interesting to see if the real circuit in any way matches up with the sim results!
Attached Images
File Type: png ODA power supply circuit.png (36.0 KB, 513 views)
File Type: png ODA power supply layout.png (39.0 KB, 503 views)
File Type: jpg CLC filter low Q 0.5R circuit.jpg (61.5 KB, 493 views)
File Type: jpg CLC filter higher Q plot.jpg (39.4 KB, 475 views)
File Type: jpg CLC filter low Q 0.5R plot.jpg (39.8 KB, 455 views)

Last edited by agdr; 27th February 2013 at 01:01 AM.
  Reply With Quote
Old 28th February 2013, 03:54 PM   #52
diyAudio Member
 
Join Date: May 2011
Location: Silicon Valley
What is good about the bootstrap connection? How is it better to connect R3 to the LT1963's output, rather than connecting R3 to ground (with suitable R value choice)?

By definition, ground has the least noise of any signal on the board. Connecting R3 to ground injects less noise into the LM317, than connecting R3 to the LT1963's output.

Suppose that an unwanted signal ("noise") does somehow appear on the LT1963's output. R3 and C18 faithfully transmit this signal to the LM317, which faithfully reproduces the signal* on its output -- the input to the LT1963. Oh great, now the LT1963 has more noise on its input, to attenuate. Why is this a good thing?

*inverted -- since the ADJ pin is the inverting input of the LM317's error amplifier, see drawing below
Attached Images
File Type: png Image1.png (19.8 KB, 102 views)
  Reply With Quote
Old 28th February 2013, 06:00 PM   #53
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
I think the big claim to fame of the tracking pre-regulator is maintaining the constant voltage across the downstream regulator. Figure 5 in the datasheet also has that arrangement: http://www.ti.com/lit/ds/symlink/lm317.pdf (opens PDF)

Excellent point on the noise. You are quite right from what I can tell, in the reverse direction noise would be conducted to the pre-regulator adjust pin. Probably not a good thing. This would apply to the Jung super-regulator too, wired up the same way.

It would be really interesting to make some measurements on the actual circuit with noise injected at the output and see what happens with that loop through the pre-regulator.

Last edited by agdr; 28th February 2013 at 06:05 PM.
  Reply With Quote
Old 28th February 2013, 07:23 PM   #54
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
Hmmm... some additional thought... maybe what needs to happen is C18 connected back to ground the standard way rather than to the output of the LT1963A. The whole point of that adj pin bypass is to keep ripple off the LM317 adj pin. Running it to the LT output would just conduct noise backwards, exactly like you say. Running it back to ground should help bypass the noise. As I recall the Jung regulator, at least in the initial paper, didn't include that cap in the tracking regulator portion. Good find!
  Reply With Quote
Old 28th February 2013, 08:12 PM   #55
diyAudio Member
 
Join Date: May 2011
Location: Silicon Valley
Quote:
I think the big claim to fame of the tracking pre-regulator is maintaining the constant voltage across the downstream regulator
If the downstream regulator does its job, its output voltage is constant. If the upstream regulator does its job, the upstream output voltage is constant too. So the downstream regulator has constant output, constant input: voila, the voltage across the downstream regulator is constant. Without any tracking / feedback connection.

You want the LT1963's input voltage to be 17.5 volts. You can accomplish this by changing R3 to 3.16K, then connecting both R3 and C18 to ground. What aspect of circuit performance is improved, if you connect R3 to the downstream output instead? What is the benefit? If there isn't a benefit, why do it?
  Reply With Quote
Old 28th February 2013, 08:54 PM   #56
jcx is online now jcx  United States
diyAudio Member
 
Join Date: Feb 2003
Location: ..
bootstrapping can have stabiiity cosequences - often not modeled right in Spice

I would use rather the series regs indpendently refed to gnd
  Reply With Quote
Old 1st March 2013, 01:13 AM   #57
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
OK, OK, I'm convinced, lol. Seriously though, thanks to both you guys for the input. Much appreciated. Interestingly enough the bootstrap reg method with the 10uF from adj to ground did lower ripple from the first reg in the sim by an entire order of magnitude, from 2mV to 200uV. Both numbers being too small to be believeable in a sim, I know, but the order of magnitude thing probably has some general merit.

But... with the series reg method, second sim circuit below, the results are the same (good) 10x reduced ripple out of the first reg, down to 200mV ripple. I would agree with you both, after some further thought here, that the series method is probably going to be less of a real-world stability challenge. jcx- I definitely agree that that Spice is going to be of limited use here in detecting stability issues. A good design upfront is the best defense. In the sim and layout I've used transistormarkj's 3.16k's.

The first spice circuit below is the (now discarded) bootstrap reg method. Second is the series reg. I've put in parameters now for a better-defined transformer, a WAU24-1000 or WAU24-1800 with about 2.4R secondary. This weekend I'll see if I can stuff in the full transformer coupled inductor model for fun.

Also going to see if I can fit in SMD pads on the bottom of the board under the rectifier diodes for optional series resistor+cap snubbers.
Attached Images
File Type: jpg ODA PS bootstrap regs spice circuit.jpg (124.0 KB, 174 views)
File Type: jpg ODA PS series regs spice circuit.jpg (141.9 KB, 180 views)
File Type: jpg ODA PS series regs plot.jpg (60.3 KB, 139 views)
File Type: jpg ODA PS series regs plot zoom 1.jpg (42.3 KB, 44 views)
File Type: jpg ODA PS series regs plot zoom 2.jpg (41.8 KB, 42 views)
File Type: png ODA PS circuit.png (35.2 KB, 114 views)
File Type: png ODA PS layout.png (38.8 KB, 116 views)

Last edited by agdr; 1st March 2013 at 01:32 AM.
  Reply With Quote
Old 1st March 2013, 12:56 PM   #58
diyAudio Member
 
Join Date: May 2011
Location: Silicon Valley
Quote:
Originally Posted by agdr View Post
Also going to see if I can fit in SMD pads on the bottom of the board under the rectifier diodes for optional series resistor+cap snubbers.
Oscillatory ringing occurs when the rectifier diode cuts off; it's a parallel LC circuit consisting of the transformer's secondary leakage inductance, and its secondary capacitance (in parallel with the rectifier's junction capacitance). The ringing can be damped ("snubbed") by adding a third parallel leg, a resistance, forming a parallel RLC circuit. A wise choice of R gives a well-damped waveform with (0.5 < zeta < 1.2) and no ringing.

This means that the snubber is placed in parallel with the transformer secondary, as shown in Figure 7 of Hagerman's paper http://www.hagtech.com/pdf/snubber.pdf . That figure's snubber is (Rs, Cs, Cx) and it is in parallel with the transformer secondary (Lt, Ct).

Remember, the rectifier is merely the agent provocateur, the initiator, the wooden hammer that whacks the bell. The bell is what rings, and the bell is what needs damping.

As Hagerman shows, this can be simulated. After measuring your transformer's secondary leakage inductance Lt and its secondary interwinding capacitance Ct, and after consulting your rectifier's data sheet to find its zero bias junction capacitance, you can build an LTSPICE model and experiment with snubbing. An example from a lab bench power supply I just built, using a TRIAD VPS36-2200 VPS36-2200 Triad Magnetics | Mouser transformer, is attached.

Compare the waveforms to Hagerman's Figures 9 and 10.
Attached Images
File Type: png before.png (32.0 KB, 117 views)
File Type: png after.png (36.7 KB, 108 views)
  Reply With Quote
Old 1st March 2013, 01:28 PM   #59
diyAudio Member
 
jackinnj's Avatar
 
Join Date: Apr 2002
Location: Llanddewi Brefi, NJ
You can improve the performance of the LT circuit by detaching the sense resistor from the regulator OUT pin and probing directly at the load. See the data sheet diagram for Kelvin connection.
  Reply With Quote
Old 2nd March 2013, 02:16 PM   #60
agdr is offline agdr  United States
diyAudio Member
 
agdr's Avatar
 
Join Date: Sep 2010
transistormarkj - sounds good! Across the secondary they go. I think that will be an easier layout job too.

jackinnj - yeah, I've been pondering that Kelvin connection. That figure is for the "fixed" voltage version of the LT1963A though, where it looks like they rebrand the "adj" as "sense". With the adjustable reg I wonder if the equivalent thing is as figure 3 here http://cds.linear.com/docs/en/datasheet/lt0117.pdf for the LT317, with the output-to-adj resistor close to the reg but adj-to-ground resistor returned to remote ground?

From past experience I better make that a "negative option" connection where sense is normally connected to the output and something has to be cut/switched/removed to remote it. I spent several hours with the first switching supply I ever used, years ago, trying to get the thing to work. Eventually discovered a sense connection needed to be connected, but no markings/instructions/manual.
  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
What is a better alternative for 317 337? Sunsun22 Power Supplies 14 18th March 2011 09:38 PM
Mathematically challenged help for lm 317-337 semi932 Power Supplies 18 5th January 2011 10:16 AM
317/337 with current boost jaudio Power Supplies 2 19th January 2007 06:42 PM
Troubleshooting 317/337 Regulator Circuits TerryC Parts 5 12th August 2006 04:42 PM
needed 317/337 double regulation jaudio Power Supplies 17 19th June 2006 07:24 PM


New To Site? Need Help?

All times are GMT. The time now is 09:56 PM.


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