0-30V Stabilized Power Supply

[redwire]

I took the liberty of making gerbers using Elecrow's cam file if anyone wants to have these made super cheap.  You can get 10 boards made for around $20 including shipping.  I'm confident in the design of the board and redwire is sure the soft start should take care of the problem with the startup transient.  I'll be doing testing on this sometime this week and report the results with pics.  If you do have this made, report back on how it works for you.  If you want to test without the soft start, just jumper the TIP141 header from the collector to the emitter and leave out the 12K resistor.  If you want to test without D10 and R15, leave out D10 and use a jumper in place of R15.  The Q1 circuit seems to now be required so leave that in.

In addition to the edit in my previous post, I redid the gerbers and I made a new schematic with the added parts.

Hold off on populating the soft start.  I started testing last night and the arrangement and values will likely need to be tweaked.    As drawn, the PS kicked on at the same time as the main cap -no good.  I then rearranged the set up by connecting the 47uf cap directly to ground and used  a 39k resistor connected between the collector and cap.  I then used a 18k resistor to the base of the TIP141.  Timing seemed pretty good (about 1 sec delay to reach 12V) with reasonable ramp up on the scope.  I did not test it with a load.  After running through the numbers (39k and 18K resistors) and the data sheet for the TIP, I am concerned that the base current of approximately  0.7mA to the TIP may be a little low under a 3-5 Amp load.  I may need to change up the resistors to increase the base current to the TIP without shortening the delay period.  Otherwise I may need to increase the value of the cap.    Will keep you posted.

 

[liquibyte]

Hold off on populating the soft start.  I started testing last night and the arrangement and values will likely need to be tweaked.    As drawn, the PS kicked on at the same time as the main cap -no good.  I then rearranged the set up by connecting the 47uf cap directly to ground and used  a 39k resistor connected between the collector and cap.  I then used a 18k resistor to the base of the TIP141.  Timing seemed pretty good (about 1 sec delay to reach 12V) with reasonable ramp up on the scope.  I did not test it with a load.  After running through the numbers (39k and 18K resistors) and the data sheet for the TIP, I am concerned that the base current of approximately  0.7mA to the TIP may be a little low under a 3-5 Amp load.  I may need to change up the resistors to increase the base current to the TIP without shortening the delay period.  Otherwise I may need to increase the value of the cap.    Will keep you posted.

Something like this?  The transistor showed up today so I'll be able to do a bit of testing as well, thanks again.

 

[redwire]

Yes that's it.    I want to try with a low value resistor and no resistor where you show the 18K reisistor to see if the delay remains approximately the same.  For the test I am using a 10,000uf main cap.  Glad you got the part.

 

[redwire]

After some experimenting, I'm not thrilled with the soft start results.  I don't know if adding all those parts are worth the benefit.

 

[liquibyte]

After some experimenting, I'm not thrilled with the soft start results.  I don't know if adding all those parts are worth the benefit.

What kinds of results are you getting?  I thought you had a delay going on before and just needed to tweak the time a bit.

I've been looking at solutions using basic transistors and there seem to be limited options there.  Is it maybe because it's not an ideal solution?  My thinking on this is that we should be using the transistor as a switch and not try to make it slowly ramp the voltage.  Isn't fast turn on ok as long as it happens after initial power is applied to the transformer, rectifier, filter cap circuit by at least the time domain that the spike lives in?  How would you delay the voltage to the base so the transistor stays off for say 200 milliseconds or so?  From what I've been reading mosfets seem to be the going rage for inrush current limiting so wouldn't this be adaptable to what we're trying to accomplish?  Something from this app note perhaps?  Figures 5a and 5b on page 9 look interesting but reading the calculations the whole thing seems to be way over my head.  The pic is from something that has to do with the usb spec and gets posted a lot.

 

[redwire]

Attached is a picture of the charge on the main cap vs output to a 12.4 V setting under a 0.5 Amp load.  The turn on is a flatter curve and slower but  I don't know what I saw yesterday when I thought I had a 1 sec delay.  I must have misread the time period.  I agree that the turn on after the cap charges is the ultimate goal and using a mosfet has many advantages. 

View attachment 42271

 

[liquibyte]

I just did a more thorough reading of that app note and I think this may be the way to go here.  I'll try and work through the calculations tomorrow and see what I can come up with.  The circuit in 5b makes a lot of sense after reading through the whole thing.  I'm sure this kind of thing came about due to switching supplies but I don't see why it couldn't be adapted to solve our problem.

 

[liquibyte]

Can someone who might know better tell me if this would work and if not, why?

 

[audioguru2]

Can someone who might know better tell me if this would work and if not, why?

I think there is a certain supply voltage where the opamps in the project suddenly begin working, maybe with severe overshoot.
maybe why most audio amplifiers do not have soft start. Instead they use a relay for delayed start so that the circuit can stabilize before it has an output.

 

[liquibyte]

I've measured the spike before and without the power supply attached so I'm thinking that even though the circuit is handling things to a certain level it's just not enough and may eventually damage the op amps themselves by being continuously subjected to +60V at power on.

If I put this before the op amps and if f I have a VGS(th) of 3.5-5.5V such as with this one, then at around 850ms or so this would turn on with the above circuit given the values of 100K and 100uF according to the calculation that I got here.  My settled voltage after rectification and filtering is 43.7V and is consistent between two separate power supplies.  So, given that value, the 100K resistor, the 100uF cap, and the equation 43.7*(1-2.71828^(-85/100000*100)) I get 3.56V @ 849.7476ms.  I made sure I verified the calculation and didn't just rely on the page itself to do it right so I'm confident in the values but I'm not sure how well it would work as I've never tried something like this before.

I don't have the mosfet I linked but I do have quite a few laying around to try this out with but am wondering if I'm just wasting my time with such a simple solution.  Should I be looking into the app note above and calculating out a circuit for either figure 5a or 5b and trying that instead?

 

[redwire]

Ok I've gotten something that seems to be working fairly well  I have a 39k connect to the high side, then a 2.2k connected to the 39k then a 47uf connected between the 2.2k and ground.    Then I have 4K7 resistor connected at the voltage divider between the 39K and 2.2k.  The other end of the 4k7 is connected to the Base of the TIP141.  Start up is nice  and shutdown occurs nearly instantly .  The first picture is startup the second is shutdown.  The blue line is connected to the main cap (10,000uf) and the yellow is the output.  Test was taken with about a 0.5amp bulb.

View attachment 42273

View attachment 42274

 

[liquibyte]

Like this?  Took me a minute to visualize that.

 

[redwire]

Yes, that's it.  What are you using to create your pics, Eagle?

 

[liquibyte]

Yeah, Eagle.  File > Export > Image @ 300dpi.  The resolution works well until you get to a bigger schematic and then I usually take it down to 150dpi otherwise it ends up huge.

I'm having a discussion over at eev about the mosfet schematic.  I think we have found two solutions.  Mine requires a zener from the source to the gate to keep VGS below 20V.  I'm going to try both and see how it goes.

 

[liquibyte]

Do you have anything that you can load this at 30V @ 3A?  Out of curiosity, I'd like to see how it behaves at the full 90W.  I'm still trying to simulate the mosfet circuit but if yours works well through the full range, I think I'll just go with that.  I'll try hooking it up tomorrow and post some results if I can get pics of it.

 

[red baron]

Hi liquibyte
I have been involved in this project for nearly two years.  I built the Picmaster version from March 09 and it worked fine for a while.  Recently while charging a 34Ah battery the power supply failed - voltage went to 42V and zero current.  What do you suspect has blown?

 

[liquibyte]

Hi liquibyte
I have been involved in this project for nearly two years.  I built the Picmaster version from March 09 and it worked fine for a while.  Recently while charging a 34Ah battery the power supply failed - voltage went to 42V and zero current.  What do you suspect has blown?

I split your post off from that thread because I wanted that one to be a showcase for people to post their completed projects.  I don't want it to grow into a discussion about it because we have two huge assed threads for that now.  I'd be tempted to merge them but since they're so long and have been going for so long it wouldn't make much sense now.

As for your question, I'm not sure but maybe someone might chime in for what to test for better than I can.  Have you done any testing?  You built it so you should know its quirks better than anyone.  I'm going to take a guess that it's either the BD139 or one of the op amps (U2?) because the voltage went to the full voltage of the rectifier/filter circuit.  I could be wrong so wait for a few more opinions, there's folks here that know way more than I do.

 

[audioguru2]

Hi Red Baron,
I have Picmaster's schematic, revision 2.10 dated 10/10/2010, which is almost the same as mine. It uses TIP3055 output transistors instead of 2N3055 ones. The TIP3055 has an epoxy case, not metal and is rated at 90W max instead of 115W max so one might be shorted C to E due to excessive heat or maybe moisture from humidity got through the epoxy.
If the BD139 or the voltage regulation opamp shorted then the output voltage would be less than the unregulated positive supply. 

 

[liquibyte]

If the BD139 or the voltage regulation opamp shorted then the output voltage would be less than the unregulated positive supply. 

Now that I think about it, that does make sense.  Having a shorted C-E on one of the pass transistors would allow the full unregulated voltage.  Meh, I'm working on three hours sleep.

 

[redwire]

Hi liquibyte
I have been involved in this project for nearly two years.  I built the Picmaster version from March 09 and it worked fine for a while.  Recently while charging a 34Ah battery the power supply failed - voltage went to 42V and zero current.  What do you suspect has blown?

A 34Ah battery?  Was it for a riding lawn mower?  If you are going to use the power supply at its limits for long periods of time you may want to consider upgrading to a couple of 2N5886 or MJ11016 output transistors mounted on a large heat sink with a small fan blowing across it.  A  2N3055 with a case temperature of 125 C  is only good for about 50 W.  A 2N5886 at 125 C is good for about  82 W.  For a 30v,  3 amp power supply operating at very low voltage and high current, the output transistor may consume 115 W.  Two 2N3055's are borderline depending on the cooling.