0-18V Pos/Neg 1A Current Limited

[liquibyte]

I have come up with what seems a fairly decent power supply using a minimal of parts that are easily obtainable and cheap.  The idea is to have a power supply for working with dual supply op amps.  The schematic in the picture differs from the LTSpiceIV simulation file for clarity.  I've got this sitting on a breadboard and it works but I've only loaded it down to 200mA so far due to my not having had boards made yet and the transistors not being on heatsinks.  Once I've got this completed I'll put up a BOM with everything I've used including the case, meters, heatsinks etc.

Updated to reflect reality.  I'm probably going to just put in analog meters for this project.

 

0-18V-Independently-Variable.zip

Edited September 22, 2015 by liquibyte
Update

[liquibyte]

I hate this new forum software. I just did an edit and it lost its mind. I had forgotten to add that you will have to install http://ltwiki.org/files/LtSpiceIV_Plus_12_2009.exe for the extra parts I used. I'm not sure if the OP07 comes with LTSpiceIV by default or not but I do know the bridge rectifier I used doesn't. If you don't want to install them, just replace the rectifier with 4 diodes. I usually use the MUR460's.

[admin]

What problem did you experience while editing the post?

[liquibyte]

I hit the submit button and I was taken back to the post but it was if I hadn't done anything in the first place, no changes.

[admin]

It seems like a JavaScript issue. What brower/operating system are you using?

[liquibyte]

Then I'd have to say that the software switch was a bad idea. It shouldn't take javascript to post, ever. I have it blocked by default on most sites unless I whitelist it and I hadn't done that yet here because things changed and I'm personally wary of things changing just for the sake of change. I think that this is probably going to have me going back to anonymous lurker status. To be honest, there's not much in relevant content here anymore and I don't see it being brought back to life any time soon without new content. It's one of the reasons I came up with this supply but since the software can't be used without arbitrary and unnecessary js usage, I don't think I want to bother with it anymore.

[audioguru]

I cannot make an attachment here with this new software when I had Windows 8.1 and Internet Explorer 11 and now I also cannot on Windows 10 and Edge or with Internet Explorer 11 even when I activated the latest JavaScript that I never needed to use on any other site.

Maybe this new software works only from a Macintosh operating system? 

[liquibyte]

Maybe this new software works only from a Macintosh operating system? 

I don't know about function but it certainly seems like it's trying to look like it.  I get why he wants everything integrated and looking the same but I despise the web 2.0 look.  The first person to do this was doing something unique, everyone else, not so much.

[CBETO3AP]

Hello liquibyte
I propose a small change that would improve the stability under variable load.

Svetozar

[audioguru]

C11 and C12 slow down the outputs too much, causing poor voltage regulation if the load current changes of if the load current is modulated. The circuit will have serious overshoots and might oscillate.

[liquibyte]

You guys are late to the conversation.  I've had the boards made and from my testing things work rather nicely.  I did make a mistake on the gerbers and forgot to connect D2 to R2 but since it was a short hop, I left the lead on the zener long, folded it over and soldered it to the lead of the resistor on the bottom.

 

I have no way to vary the current dynamically and I'd not imagine anyone doing that so I'd say, all in all, things work as expected with no oscillation at all.  I've got less than 20mV ripple at full load on the output.  For what I designed this for, I'd say it works nicely.

[CBETO3AP]

I am glad you are happy with what has been achieved. I did not mean to criticize, simply insignificant change without complicating the circuit can eliminate the fall of 0.62v when consumption is 1A. The audioguru also right, C11 and C12 better be around 0.XXXuF.

Without wishing to disturb: Svetozar

[liquibyte]

I am glad you are happy with what has been achieved. I did not mean to criticize, simply insignificant change without complicating the circuit can eliminate the fall of 0.62v when consumption is 1A. The audioguru also right, C11 and C12 better be around 0.XXXuF.

Without wishing to disturb: Svetozar

Your change adds significant oscillations to my simulation output on the positive side of things.

C11 and C12 slow down the outputs too much, causing poor voltage regulation if the load current changes of if the load current is modulated. The circuit will have serious overshoots and might oscillate.

As for the 2.2u cap, it could have just as well been 1uF but 2.2uF is what I had on hand that solved the issue I was working on also in regards to oscillations.  I got the value from doing some research into problems I was having and this was a recommended solution and, no offence, it worked.  I'd link what I was reading but I can't find it again.  Suffice to say that it was from a reputable source and not instructables or the like.  I think it was actually an ap note and I'll link it if I find it.

 

The bottom line is that while math and theory can go a long way, up until you build something you never know what you're going to get.  Oh, and always run board checks.  The missing trace was a dumb mistake on my part.  The pic of the board shows the fix I had to do.

 

This is a redesign of something I found online and wanted to see if I could simulate it an manipulate it to do what I wanted.  My changes worked so well that I wanted to put it on a manufactured board so I did.

[CBETO3AP]

Your change adds significant oscillations to my simulation output on the positive side of things.

 

 

C11 and C12 slow down the outputs too much, causing poor voltage regulation if the load current changes of if the load current is modulated. The circuit will have serious overshoots and might oscillate.

If you like to experiment try This

"I have come up with what seems a fairly decent power supply using a minimal of parts that are easily obtainable and cheap.  The idea is to have a power supply for working with dual supply op amps."

A very basic and stable example. The op-amp burn equal to 1A or 5A.

Very good work

 

Edited October 28, 2015 by CBETO3AP

[audioguru]

The two series diodes at the outputs of the regulators RUINS their excellent voltage regulation. Why two diodes, why not just one diode, it will also ruin the voltage regulation but not as much as two diodes?

 

The sliders on your 5k pots are shown disconnected.

The 240 ohm resistors are for the more expensive LM138 and LM137. LM338 AND LM337 need 120 ohms (and the pot value also halved) to prevent the output voltages from rising if the load current is low.

[CBETO3AP]

Hihi because I'm lazy. I just grab the first Internet scheme as an example. I do not know who and why has put diodes. The potentiometer is understood. And of course, LM317, LM337 are more appropriate. It's just as basic example.

[liquibyte]

The fact that you even suggested that circuit tells me that you either know nothing and are trying to fake it or that what you think you know is entirely wrong.  Linear regulators as a lab supply is just stupid and I'm tired of seeing that dumbass circuit all over the net being suggested as so.

[CBETO3AP]

If I'm wrong about something (of course this can be and prefer to see in the community), there is audioguru, who always find the error and tell, and always right, which I respect a lot. This is already on frame of contecst so I would like to haver not said anything about your "work of art". Practically not get anything more than what can be done with the LM317, LM337. In fact you've gotten worse stability, thereby constructively we have discussed before. I will not bother to draw a circuit totally different from yours, it's easier to grab any circulating on the internet and we would have the same result as minimum. Everything I've said has been with good intentions, so far. I am sorry that you've had that bad.

[liquibyte]

In fact you've gotten worse stability, thereby constructively we have discussed before.

Really, 20mV p/p is worse stability?  I've got a nice scope trace I'd like you to look at then.

[CBETO3AP]

The truth is that I respond with fear, you're very reluctant to my comments and I need not bother you.
For me it is obvious that feedback must be in the output of the circuit. I suspected but was not so obvious that the C11, C12 are very large but audioguro corrected me and then I saw that could cause the problem.
  You need to know the why of each component in your circuit to reason and improve it. This is sometimes quite difficult, especially when you copied the outline of another without understanding. I'm not electronics guru, so I entered this community, to continue to learn from those who know more. Here I ended the conversation with you, you obviously do not need my help.

[liquibyte]

You can do a search for output decoupling caps to ground and you will find several references to 1uF to 2.2uF for the values in an effort to solve oscillation.  I used these because of that actual problem I was having.  I do understand each component in the circuit having studied it extensively and having modelled it in LTSpiceIV extensively to arrive at the circuit that I posted.  Bringing in linear regulators to the circuit added nothing to the conversation whatsoever.  I did not copy that other circuit, I based mine off of that configuration.  By the way, the 0-30V circuit uses the same principles and quite a bit of the same configuration.  I don't know if you've noticed but I posted a modified version of that circuit that I'm working on as well.  I removed the part of the circuit that was the negative regulator because it really added nothing to the circuit and introduced issues that has made it hard to fix.  We now have a 10V reference and a 0.1 ohm current sense resistor that suffers from much less heat than before.  Still think I don't understand what's going on?

[audioguru]

I am glad that the value of the R7 heater has been reduced to 0.1 ohm. The original was 0.47 ohm which heated with (3A squared x 0.47 ohm=) 4.23W. The TL081 opamps that were used had a maximum input offset voltage of 15mV (!) so the minimum regulated current with an accuracy of 20% was 15mV/0.47 ohm x 1/5= 6mA but they claimed 2mA.

I do not think anybody needs current regulation less than 20mA (for an LED?).

[liquibyte]

I am glad that the value of the R7 heater has been reduced to 0.1 ohm. The original was 0.47 ohm which heated with (3A squared x 0.47 ohm=) 4.23W. The TL081 opamps that were used had a maximum input offset voltage of 15mV (!) so the minimum regulated current with an accuracy of 20% was 15mV/0.47 ohm x 1/5= 6mA but they claimed 2mA.

I do not think anybody needs current regulation less than 20mA (for an LED?).

I tried my best to fix a few of the more glaring issues that I had while I was testing.  The one thing that always bothered me was the current sense heater.  I felt like I should be using this as an on demand water heater.  I'm actively working on prototyping out the changes to see what works and what doesn't.  The one thing we lose is getting voltages close to 0V but I don't think anyone really needs that either.  I haven't looked at different op amp solutions yet because the TLE214x family worked so well but my guess is that a nice rail-to-rail could go in there now without issues due to the 36V regulator circuit and come close to addressing that issue.  My focus was on the transient issues and a better current sensing scheme.  I'm planning a voltmeter/ammeter for this using modern parts.  12bit ADC, Atmel microcontroller (up for debate but I've got some blank 328P-PU's), and a MAX7219 to drive the displays, and a 4-wire sense resistor.  Doing the meter is more for the educational value than anything else for me but some folks might find it useful so I plan on releasing everything all at once when testing is complete.