Home made dual rail power supply problems

[joedevola]

Hello everyone.

I am a relative newb to electronics.  I like to build audio gear (amps speakers etc), have built half a dozen scratch built tube amps, active crossovers etc.  My 16 year old son has started to develop an interest in electronics (more mechatronics actually) so for his birthday I bought him some components, a good multimeter, a second hand scope and parts to build a dual rail lab power supply.  I used the schematic from Rod Elliot's site (Elliot Sound Products) here http://www.extremecircuits.net/2010/05/adjustable-current-limit-for-dual-power.html.  It is a pretty basic LM317/337 dual rail supply with a 24-0-24 transformer to give a variable dual rail supply up to about 25 volts.  That part of the plan worked.  Next I saw this schematic http://www.extremecircuits.net/2010/05/adjustable-current-limit-for-dual-power.html for a circuit to limit the current.  For the life of me I cannot get the current limit to work.  I have built it on perf board for the supply and it did not work, then I tried building it on a breadboard and got the same results.  Could someone check the current limit schematic to see if there are any errors and/or suggest a simple implementation of a current limit for a dual rail supply.

Complicating the issue is we have built the enclosure, have drilled a hole in the front face for the current limit pot and have stencilled in a label for it so simply not implementing the current limit would leave the supply with an ugly hole and also teach my son that to give up on a project is an acceptable outcome.

Any help appreciated!

Thanks,

Chris

[w2aew]

I think the basic schematic looks OK.  One thing to be careful of is that the metal tab on most of these types of 3-terminal regulators (on the TO-220 package), including the LM317, is electrically connected to the Vout pin.  So, do NOT attach the the various regulators to a common electrically-conductive heatsink. 

If you can describe what isn't working, and some voltage measurements that you are getting while it is mis-behaving, maybe we can help you figure out what is going on.

Alan

[joedevola]

Thanks for taking the time to respond.  I am about to step out for 3 days with work.  Will report voltages when I return.  Yes, tabs on TO220 devices are electrically isolated...  Voltage rails at filter caps prior to connection to regs is +/- 34 volts.

Cheers,

Chris

[joedevola]

Thanks guys,

Still away with work at the moment, but if someone could help with an explanation of the circuit in question, it would really aid by ability to troubleshoot when I return.

Specifically, I am confused as to what specifically is happening with the voltages and currents around R1/R3 and R2/R5 and what the LM317/337 is trying to do with regard to the 1.25 volt reference.

From memory, I could get IC1 and IC2 to work independent of the IC3 circuitry (i.e. with just a 1.2R sense resistor it would limit current to 1 Amp).  I could also get IC3 to work independently, in that I could get it to limit from 10 to 60 mA when connected to a 10 volt supply.  From memory, I don't think I could get it working when connected across the rails, that is input +34 volts to output on -34 volt rail.  I see the data sheet has a 40 volt limit from in to out, but the original schematic call for +/- rails of "at least 22volts" and my supply is +/- 34, maybe a problem here?



Values:
R1, R2  1.2R
R3, R5  120R
R4  22R
P1  100R

Thanks for the help,

Chris

[joedevola]

To me it looks like the IC2 setting 10-60mA will provide a bias voltage to the Adj pin of IC1 and 2 by the voltage drop across R3 and R5.  As they are specified as 120R, that would give an adjustment of 1.2-7.2 volts.  I would have thought 0-1.125 volts would have been the appropriate bias adjustment?

If my thinking is right, perhaps this amended schematic might work?



R1 and R2 are 1.2R to limit output to 1A. IC2 sets a current of about 15mA with R3 at 82R. R5 and R6 each drop about 15 volts to keep the voltage across IC2 at about 34 volts (+/- rails are +/- 34 volts) (or perhaps 15v zeners might be more appropriate here?). LEDs drop about 1.7 volts, so voltage divider dual potentiometer (R4) can set a bias of 0-1.7 volts to the adjust pins of the LM317/337 IC1/3.

Am I on the right track?

Thanks for the help so far.

Regards,

Chris

[jimmc]

I can see a few problems with the original circuit.
Firstly as you say, IC3 sees almost the full voltage between the + and - inputs limiting these to +20V max.

Secondly the values around IC3 on the original circuit look wrong.
The current limit  I_L is given by (V_REF - (I_IC3 x R3))/R1 = 1.2- (I_IC3 x 120)/1.2 where  I_IC3 is the current through IC3
Thus to vary the current limit between 0 and 1 A, I_IC3 should be set from 0 to 10mA.
Now I_IC3 =1.2v/(R4+P1)= 1.2/22 max = 54mA and 1.2/(100+22) min = 8.2mA which is far too high!
(The minimum current of the LM317 is up to 5mA so this is never going to work too well)

A third problem is the lack of reversed biased diodes across C4 and C5, without these a load between the + and - outputs could cause whichever of the rails limits first to reverse polarity.


Your circuit is more sensible, a couple of minor points...

1/ The connection between the end of each pot and the adjust terminal should not be there.
2/ R4A is backwards, clockwise rotation should move the wiper towards R1 (R4B is correct)
3/ Add diodes from the outputs to 0v to protect against polarity reversal.

One final thought, replacing the LEDs with bandgap references such as the LM385Z would give better stability.
In fact IC2, R4 and R5 could be replaced by a single resistor of around 22k (1.2mA through the pots and 2mA through the LM385Zs)

Jim

[joedevola]

Thanks Jim!

It can be disheartening when starting to learn something new when projects do not work.  I guess I learned a lesson that not all info on the web is good

Is this something like what you suggest?




The polarity protection diodes are on the project, but at the outputs after the voltage regulator section (not shown).

I will have to order the parts (LM385), but have to order other stuff anyway, so no big deal.  Will breadboard first to make sure it works OK and report back in a few days.

Thanks again!

Chris

P.S. Edited for change to schematic

Did not have any LM385, so breadboarded the LED/current source schematic above, but with the changes to bias wiring as picked up by Jim and it now works very well (changed current for bias less to 10mA and used 2 x 1.5k resistors to limit the voltage across the bias LM317)

Thanks for the help, project back on the rails and I have learned something!

Chris

[jimmc]

Hi Chris ,

Yes, that's what I was thinking of, however I've realised there is a tolerance problem.
Not all advice on the forums is good either.

If  the 1.2V version of the LM385is used the worst case minimum current limit is around 80mA
(LM317 V_REF at 1.3V, and LM385 at 1.205V). With typical values it works, but you can't design that way.

It would be better to use the 2.5V version with a resistor between the LM385 and the pot (820R nominal), this should allow the minimum current limit to be set to zero. (without the 820R, zero current limit would occur at around the mid point of the pot 'wasting' half the travel.)

Both pots are now the wrong way round, assuming you want the current limit to increase with clockwise rotation. (arrow indicates clockwise rotation)

I hope I haven't disheartened you further

Jim

ps  If the LED circuit is working then why not stick with it (again a resistor between the LED and the pot would increase the arc of the pot used, 330 to 470R depending on the LED voltage).
My only concerns were that the tempco of the LED forward voltage is around -2mV/^oC and that  parameters are not well defined.

[FenderBender]

Hey Joe and Jim. I really like the progress with this supply. I was just able to order the parts for my own build. I was going to use a dual rotary switch to dial in voltages, but I really like your stuff here.

I looked at that LM317 current regulator also, and it confused me, though I didn't build it. Looks like a better approach here.

Can you post a final schematic for reference when you are done?

Do you think you should add any way to calibrate the tracking since a dual pots might not track exactly the same over time? A trimmer in series or something?

[joedevola]

Thanks Guys,

Yes, when I breadboarded I had the pot the wrong way around, Sod's Law.  I will post here what is on the breadboard and working:



The LED drops about 1.8 volts, so there is a little dead space in the pot adjusting up from zero current.  I thought about a resistor to adjust, but the dead space is not too bad.  5k pot used, as that was all I had in the parts bin.  Adjusted the current of the bias LM317 to 10mA nominal (9 measured).

Idea for LEDs for bias reference was from valve amp experience using them to set cathode bias.  I used LED also as a voltage reference in my last amp to trim bias the second grid in a differential pair that was DC coupled to the next stage to get equal output voltages (had much help with that from guys at DIYAudio.com).

For Fenderbender, I will draft a full schematic of the current limiter I have above integrated with the voltage supply from Elliot Sound Products.  Hope to get it done tonight.

Thanks again guys,

Chris

[joedevola]

This is how it looks all together.



I will have 1R sense resistors on the outputs so that I can use a digital panel meter to monitor/set current.

Credit for the voltage section to Rod Elliot of Elliot Sound Products, please see this link to the project and other great stuff http://sound.westhost.com/project44.htm
Hope this helps!

Cheers,

Chris

[jimmc]

Hi Chris,
Glad to see you've got it sorted.

By the way there are some 'typos' on the diagram...
As B1 is drawn TR1 should connect to top and bottom of 'diamond' C1 to RH and C2 to LH.
Both halves of R11 are backwards.
Links are missing from D3A/C3 to IC4adj and from D4A/C4 to IC5adj.

I confess that I don't like the voltage adjustment, I think you'll find it's very non linear.


To answer FenderBender,  tracking between the rails is not the best I've seen, but it doesn't matter for most applications.
(The usual method is to have one supply as a 'master' and a feedback circuit to slave the other rail to it.)

Jim

[joedevola]

Thanks Jim,

Your help has been invaluable.  I fixed the typos you mentioned, thanks for that.  I just edited the image above so there is not just a stream of incorrect schematics in the thread.

If you can suggest a better voltage control, please go ahead and make a suggestion.  I have that part of the supply hard wired already, but it would be for the benefit of any future projects or people looking at this thread.

Also, noted last night while testing that while both rails regulate the current, the output of the positive rail was reading +39 volts to common with 1 Amp, but negative rail was showing about -19 volts to common with 1 Amp.  Input to the current regs was +/- 39 volts measured to common.  Common is floating at the moment.

Cheers,

Chris

[FenderBender]

So does that schematic still need revision?

[jimmc]

Hi Chris,
On the diagram left and right connections of B1 are crossed (cathode end is positive).

The standard way to adjust the voltage is a variable resistor from adj to ground (and a fixed resistor from output to adjust).
This gives linear control but can suffer from spikes as the pot wears and the contact resistance between the wiper and track starts to increase.
Probably the best way is to use a separate voltage reference and Op-Amp as Dave has described in his power supply.

Not too sure where you were measuring the voltages.
By 'Common is floating at the moment' do you mean that you are using a single load between + and -.
If you do then the voltages you measured could be normal if you are into current limiting.
Ideally in current limit the current should be independent of the output voltage so whichever side limits first (by even the smallest amount) will drop its voltage until your load draws the set current.
The other side will not be in current limit and will maintain its output.

Jim

[fletch]

Not sure what happened to the schematics mentioned above but it sure would be helpful if someone could repost them.