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 28th May 2006, 01:38 PM   #1
Shredly is offline Shredly  United States
diyAudio Member
 
Join Date: May 2006
Default Them dang LM338s

I keep lookin at 'em, thinkin how to get two 40V rails for my power amp. One's easy; gettin the negative one is also easy if you have a dual-40V transformer, but what about a full-wave center-tapped design? What will happen if you tie the input of the negative one to the ground of the other, and the center tap? I don't want to let the smoke out... particularly not with what I'm paying for those 338s.

Suggestions? Cautions? I've seen Pedja Rogic's dual rail supply, nice job BTW, I saw a mention of CarlosFM's improvements but haven't managed to come across them so far, any pointers greatly appreciated.
  Reply With Quote
Old 29th May 2006, 06:19 AM   #2
Shredly is offline Shredly  United States
diyAudio Member
 
Join Date: May 2006
OK, so here's a hack at it. I'm still thinking about the bypass diode on VR2. I suspect that makes ground always more negative than the center tap. I'm also not entirely happy about the cap from the center tap to ground; it might store charge and push the center tap up.

Comments? Suggestions?

One additional note: I found a second source for the two-transformer dual output supply with two positive regulators... in the National LM340 dox, page 14. My first source was the NEC dox for their regulators, page 21, circuit 14. Both show diodes bypassing the negative to the positive output of each regulator- that is, from -V to G, and from G to +V. The NEC dox claim that if you don't put them there something can go negative during startup and really screw things up. I noticed that was missing from Pedja Rojic's design. I still haven't seen CarlosFM's.
Attached Images
File Type: jpg twin40v_2.jpg (55.6 KB, 440 views)
  Reply With Quote
Old 29th May 2006, 06:19 AM   #3
Shredly is offline Shredly  United States
diyAudio Member
 
Join Date: May 2006
OK, got it.

The key is to add two diodes to the center tap; one pointing from ground of the positive regulator to the center tap, and one pointing from the center tap to the positive side of the negative regulator.

A second key is to ensure that the regulators don't fall out of regulation due to light load (which will allow ripple into the load, AKA "power hum" from the quiescent amp). To do this, I've added a 1K resistor between each rail and ground. Thus, quiescent, the supply draws 40mA on each side. It can supply 5A (minus that 40mA; fine, 4.96A :P ) from either rail to ground, and 5A from rail to rail. D1 protects against the center tap going positive and dragging ground up, and D2 protects against Vin to the regulator from being dragged negative by the center tap should it go negative. CR1 supplies full-wave to each rail, and the negative regulator conditions the negative rail thru ADJ and the 10uF tantalum cap (which bypasses ADJ to AC ground and gives ripple rejection of 75dB).

A major factor in the design in order to get the best possible noise performance and load regulation is to ensure that the bottom of R2 is connected to ground where it enters the load, rather than on the power supply board; however, both ends of R1 must be very close to the regulator. C1 and C2 provide for sudden demands from the regulators, and C3 and C4 decouple them from the power line. C7 and C8 provide "clean" power for sudden demands from the load, and C9 and C10 again decouple the regulator (they are actually on the load board, it was using unregulated power which is why I'm doing this in the first place :P ). I'll probably add another pair of .1uF disks to the outputs of the regulators, just for grins, since C9 and C10 are so far away.

Finally, D3 and D4 provide a discharge path for C7 and C8, so they don't destroy the regulator at startup or if there is a line short, and D5 and D6 send the discharge from C5 and C6 to the same place, since these caps are dangerous to the regulator in the event of either load or line shorts. C1-C4 cannot damage the regulators because they are on the input sides.

Someone have a look and tell me I'm not insane, please.
Attached Images
File Type: jpg twin40v_3.jpg (73.9 KB, 438 views)
  Reply With Quote
Old 29th May 2006, 07:22 AM   #4
Eva is offline Eva  Spain
diyAudio Member
 
Eva's Avatar
 
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
That arrangement is inefficient in terms of diode losses and transformer usage. Furthermore, it will cause transformer saturation in case of any load imbalance between rails. A transformer with two independent secondaries is required in order to get reasonable performance.

Also, don't waste capacitors connecting them at the output of the regulator, they are required at the input for rectification instead. 100uF at the output is enough.
__________________
I use to feel like the small child in The Emperor's New Clothes tale
  Reply With Quote
Old 29th May 2006, 10:21 PM   #5
Shredly is offline Shredly  United States
diyAudio Member
 
Join Date: May 2006
Thank you for the advice.

I believe that D1 and D2 are necessary to prevent the negative regulator from browning out. Actually, it is D2 that is the main element in this regard; D1 merely preserves the symmetry of the loading from the center-tap's POV, mitigating the potential for unbalanced loading at least from the regulator components. Do you agree?

The system runs from standard US 120VAC, hot, neutral, and ground. Other than chassis heating, I am relatively uninterested in efficiency; thus, the minor inefficiencies of two 1.4V diode drops are of little concern to me in a 40V supply. Were the output voltage less, making the inefficiency greater, or were I working under extreme conditions where heating or waste of power were of great concern, I would avoid a linear regulator in the first place and ask you folks for advice on designing a SMPS; I have been working on a Cuk converter for another application for quite a while, titivating it to try to get the efficiency up in the 90th percentile (I'm an amateur astronomer, and I want extremely high efficiency DC-DC conversion to condition battery output (10A 12V) and currently it's at 87%- I may ask for advice later). In intermediate applications, I suspect your comments on efficiency should help others guide application of this supply, and I agree with them.

In regard to the output caps, they are already present on the amp board; and they present a surge current source that might keep the regulators from shutting down should the load beyond them make a sudden demand. While I could dismount them, it would be a big hassle and I have much newer technology 3300uF cans to use in the input filter. Under other circumstances, you would be absolutely correct, there is no justification for using such large output caps.

The application is a guitar amplifier, and the relevant design details are:
1. Opamps for the preamp/tone amp, and on the chorus and reverb board, powered by +/- 15V derived on the chorus board from the main +/- 40V supply; there are a few discrete transistors, serving mainly as high-gain class-A small signal amplifiers, but they are arranged to run from +/- 15V as well. I see no signs in the design of unbalanced load, and if there is any, it is on the close order of 100 milliamps, whereas the total demand is on the close order of 3A.
2. Power amp is Darlington pairs (discrete, not single-package) arranged as the main elements in a classical B-class push-pull power amplifier, with symmetrical adjustable active biasing, designed to be adjusted to place the ground of the class-B amp as close to true ground as practical. As far as I can see, the power amplifier elements are also symmetrically designed, far more so than the preamp and chorus/reverb; I see no reason to believe the power demands from the power amp will not be highly symmetrical, to within tens of milliamps or less.

The amp currently runs completely unregulated; that is, the center-tap from the transformer goes to ground, and the output of the bridge rectifier (CR1 in my schematic) goes to the two 6800uF caps at the output of my supply, and the two 0.1uF poly caps (C7-C10). That power is directly applied to the power amp without further ado. If I understand matters correctly, this presents the same potential problems with saturation that my design does; all I have done is add regulation. It's a late-'80s Ampeg SS-140C, if you're curious. And my goodness, you should see how these folks mistreated ground. But 'nuff a that, it's off-topic.

Based on this I believe that except under component failure conditions in the load or the regulator, the load will remain balanced (on average over time, that is; a pure sine wave would of course alternately demand current from one side then the other at its fundamental frequency, but the preamp design limits the bottom frequency to perhaps 40-50HZ; this is about the bottom note a guitar can make, and it is, after all, a guitar amp). I will check, by hooking the existing supply up and measuring the currents under quiescent, normal, and extreme conditions.

While I know my way around general electronics, I am not a professional; I believe that when you speak of "saturation" of the transformer, you mean that the unbalanced magnetic field will saturate and heat the core, which could damage or destroy the transformer if prolonged or extreme, but I am not certain of the details, so could you please explain in some detail so I can evaluate the risks? Also, if my understanding of the original design is correct, I believe these risks are present in that design; could you please confirm or deny?
  Reply With Quote
Old 29th May 2006, 10:21 PM   #6
Shredly is offline Shredly  United States
diyAudio Member
 
Join Date: May 2006
The transformer turns ratio, by the way, is 2:1.
  Reply With Quote
Old 29th May 2006, 11:02 PM   #7
Shredly is offline Shredly  United States
diyAudio Member
 
Join Date: May 2006
And just for clarity, I have amended the schematic slightly; first, the exact measured voltage at the output of the rectifier has been added. With a turns ratio of 2:1, one expects 60VRMS out of the secondary, and when I measure the output of the rectifier with my Fluke 23 (which measures true RMS, at the time the meter was made it was a major selling point), I get 58.8V (indicating that my local line voltage is running a couple volts over spec just now- the diode drops, you know). Taking this RMS voltage and multiplying by pi/2 I get 94.25VP for the peak voltage of the secondary; dividing by two to get the voltage from the center tap, I get 47.12VP, and this should very closely approximate the voltage after filtering with large caps. Note that there is sufficient headroom to provide for the ~3V minimum regulator brownout, even if the mains power fluctuates down by 5% (to 44.77V). As it is, the voltage at the inputs to the regulators will be 45.72VDC, including the diode drops, and at a setpoint of 40V (which is what the amp is designed for), the drop will be 5.72V across the regulators which is well within their design specs and will cause minimal derating. It is about an additional 2-3V beyond the brownout level.

Second, I have placed a dot-dash lined box around the components I propose to add; the components outside the box are existing, with the sole exception that the center tap is directly connected to ground in the existing design.

Hopefully this will clarify things for you.
Attached Images
File Type: jpg twin40v_4.jpg (91.4 KB, 346 views)
  Reply With Quote
Old 31st May 2006, 05:13 PM   #8
Shredly is offline Shredly  United States
diyAudio Member
 
Join Date: May 2006
The power amp schematic is available in the Solid State amp forum; it is worth noting that the +/-15V supply is done by a 7815/7915 pair on the chorus board, and sent also to the preamp board; this implies that the entire demand of both the preamp and chorus is under 1A for each rail, or 2A total, and more probably under 500mA. I have scrutinized things, and the demand looks very balanced, as well- certainly to within 100mA, and probably much less than that. No-load measurements show that it is balanced to within the tolerance of my meter, which is four places on the 10A setting. (I didn't dare to check it on the 300mA setting- it's a nice meter, has lasted a long time, and I'd hate to blow it up.)

If the total demand of the system exceeds the 5A on each rail that the LM338s can put out, my intention is to redesign for 10A using the opamp idea from the LM338's datasheet. I have ordered enough LM338s to do that, and for that matter to do 15A on either rail, and I have opamps on hand.
  Reply With Quote
Old 1st June 2006, 01:16 PM   #9
diyAudio Member
 
Join Date: Oct 2002
Location: Royal Oak, MI.
I wish I could be of help, but I am even much less of a designer than you!

I was wondering if you might know what the amp/voltage range of this regulator circuit would be? I'm asking because I'm looking to power several LM3875's from a transformer that is at too high of a voltage to use "as-is". I have been looking for a good circuit for it.

Thanks!

Tall Shadow
  Reply With Quote
Old 1st June 2006, 05:47 PM   #10
Shredly is offline Shredly  United States
diyAudio Member
 
Join Date: May 2006
LM338s will handle, IIRC, up to a 32-volt (see the data sheets) differential between input and output, but not without a cost: like all 3-terminal regulators, the higher this voltage, the less current they can source.

What is the voltage (or better, the turns ratio, and what is your electical utility sending you? For example, "my transformer has a 2:1 turns ratio; I live in the US and we use 60Hz at a nominal 120Vrms.") of your transformer? And what voltage do you intend to provide to your load? Do you need positive and negative rails? Is the load "balanced" between the rails if so? How much current does it need?

If you'll answer these questions, I'll tell you if this design is useful for your purposes, and advise you on how to use it (if I think you should use it at all). And I'll show you what criteria I use to judge it, as well, so you can do it for yourself next time, as well as be comfortable with my advice.
  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
REAL diy, dang nab it! DSP_Geek Everything Else 0 23rd July 2005 11:25 PM
off topic dang that's a big tank karma Everything Else 13 18th June 2005 11:39 AM
Dang these Jordans! Dave Jones Full Range 56 14th May 2005 03:53 PM
Dang... just fired up my first channel of the JLH 20w... Lisandro_P Solid State 2 1st March 2002 01:04 AM


New To Site? Need Help?

All times are GMT. The time now is 11:04 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