Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Diy Bench power supply Oscillating with Car lamps as load

Status
Not open for further replies.

diydidi

Junior Member level 1
Joined
Apr 12, 2016
Messages
16
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
156
Hi
I have been woking on a bench power supply for the past two months or so.
I decided on a discrete regulator design with adjustable voltage and and current limit.
Yesterday I boxed most of it, and decided to test using a couple of parallel connected 12V car bulbs. The bulbs were flickering slightly. Scoped the output and revealed a 16- 25HZ almost square wave/ round edges type oscillation or motor boating.
If i turn down the current limiting it gets even worse.
With my DC load all works as it should.
My electronics knowledge doesn't stretch as far as to fix this. The sense leads come together at the croc clips of the output leads. So there are four wires coming out of the box. The negative output terminal connects directly to neg of main filter caps.
Connecting a 4700uF capacitor across the leads at the load doesn't help much.

Tried the 100uf across sensor wires but no effect.
It’s just strange that if I short the + sense wire to the main + out on the pcb instead of at the croc clip, everything works great, This confirms that the current board is actually working.
Tried making C2 smaller, this too does nothing.
Something is up here....
Any ideas? See schematic attached. Also attached the output waveform.
 

Attachments

  • Paul Power supp.pdf
    48 KB · Views: 102
  • 87800-8493931496fb15c4b04849354805dcc3.jpg
    87800-8493931496fb15c4b04849354805dcc3.jpg
    3.3 KB · Views: 115

Hi,

.. A warm welcome to you at edaforum

***

try this:

connect a 100R within the case from +_Output to +V_sensing
and a 100R from 0V_Output to -0V_sensing.

This mainly prevents from high outut voltage in case one of the sense lines is disconnected externally.
With this you can use the power supply in two wire mode. (without external sense wires)
As soon as you connect the external sense wires the resistance of the connected sense wire is much lower than the 100R and the true "sense mode" is active.

Try if the problem goes away wiht the two resistors. There is a little change because of decreased (and faster) feedback.

If the problem persists, then connect a 100nF ceramics capacitor across the just installed 100R resistors (two capacitors).

****
If the problem still persists, then one needs further analysation on your complete regulation loop.

The problem with incandescent light bulbs is, then when cold (at the time when powered ON) they draw much more current than when alight. Mybe too much current.

***
Let´s try .. and see what happens.

Good luck

Klaus
 

Thanx Klaus. Will give this a shot and report back.
This is probably the best answer I've gotten from all 5 forums I'm on. Won't mention names. :) Glad I joined!
 

Incandescent lamps present a much lower resistance cold,
than hot. Are you flickering due to current limit? Can you
measure the load current in the time domain ('scope and
sense resistor)?
 

Incandescent lamps present a much lower resistance cold,
than hot. Are you flickering due to current limit? Can you
measure the load current in the time domain ('scope and
sense resistor)?
Hi Yes this is happening at max current limit 4A. The bulbs draw more than this close to 5A.
If I turn my current limit pot down more, it worsens.

- - - Updated - - -

Here is the original circuit (source unknown), The schematic I originally posted is the one I built, and is a slightly modified version of this one. Values exactly the same, except I'm using 4 paralleled TIP35C transistors.
Also notice that there is a 10K resistor missing from my original schematic. It goes between D4 anode and D17 Anode.
 

It may be a transient stability issue with your design. I see a few RC time constants, which may be responsible for this behavior.
 

I agree, without doing any math, intuitively I would guess you are activating the current overload which works to drop the output voltage, this is then reducing the current and in turn taking it out of overload condition. The time constant of the RC networks is slowing it's response and creating the LF oscillation you see.

You can prove if I'm right by TEMPORARILY wiring your 100uF capacitor across C2 or maybe C5 to see if it changes the oscillation frequency. Please don't treat the new value as a fix as it may have other consequences on the stability but if the frequency changes it will prove the point.

Brian.
 

The schematic does not make sense to me. I can't see how Q7 ever switches off since some current always seems to be flowing through the R6-D4 branch into Q7 base regardless what Q5 or Q4 does to Q6. R12 is too large to affect the current through R6. Maybe there are some errors in the schematic.
 
Last edited:

It's certainly an unusual design and there appears to be at least one error in the schematic in that the two input transformers have primaries in series instead of parallel. If it has been wired that way, there is a good chance there is insufficient internal supply voltage/current anyway.

The floating 20V supply is referenced to 10V above output voltage and 10V below it so looking at worst case load - a short circuit - the top of R6 has 10V on it, if you ignore the bias feed to Q7 from Q6 (as it should be cut off) and subtract the Zener voltage, the maximum possible base current to Q7 would be (10 - 5.6)/10K = 440uA. That also ignores the Vbe drop in the output transistor so you are right, it never switches off but it is starved of enough base current to fully conduct.

Brian.
 
The schematic does not make sense to me. I can't see how Q7 ever switches off since some current always seems to be flowing through the R6-D4 branch into Q7 base regardless what Q5 or Q4 does to Q6. R12 is too large to affect the current through R6. Maybe there are some errors in the schematic.
Hi. As I mentioned, there is a 10K resistor missing. It is connected between D4 anode and D17 Anode.
 

KLAUS. The solution you proposed works a charm!!
The capacitors alone sorted the issue, but I will add the 100ohm resistors in parallel with the caps, as you have suggested.What wattage resistors should I use?
Also why are the caps solving this issue?
Now the bulbs are fine, but when I try a 12V motor, it goes haywire until a connect a 4700uF cap across the motor terminals.
Maybe I should add a switch in the power supply internally that switches a cap across my supply wires only to be used with motors. I am open for suggestions wrt this too.
Then I would like to fix any extra design flaws in the circuit while I'm at it.
Seems like there is/ might be some judging by the other replies here.
Klaus you are a legend!!
Hi,

.. A warm welcome to you at edaforum

***

try this:

connect a 100R within the case from +_Output to +V_sensing
and a 100R from 0V_Output to -0V_sensing.

This mainly prevents from high outut voltage in case one of the sense lines is disconnected externally.
With this you can use the power supply in two wire mode. (without external sense wires)
As soon as you connect the external sense wires the resistance of the connected sense wire is much lower than the 100R and the true "sense mode" is active.

Try if the problem goes away wiht the two resistors. There is a little change because of decreased (and faster) feedback.

If the problem persists, then connect a 100nF ceramics capacitor across the just installed 100R resistors (two capacitors).

****
If the problem still persists, then one needs further analysation on your complete regulation loop.

The problem with incandescent light bulbs is, then when cold (at the time when powered ON) they draw much more current than when alight. Mybe too much current.

***
Let´s try .. and see what happens.

Good luck

Klaus
 

It's certainly an unusual design and there appears to be at least one error in the schematic in that the two input transformers have primaries in series instead of parallel. If it has been wired that way, there is a good chance there is insufficient internal supply voltage/current anyway.

The floating 20V supply is referenced to 10V above output voltage and 10V below it so looking at worst case load - a short circuit - the top of R6 has 10V on it, if you ignore the bias feed to Q7 from Q6 (as it should be cut off) and subtract the Zener voltage, the maximum possible base current to Q7 would be (10 - 5.6)/10K = 440uA. That also ignores the Vbe drop in the output transistor so you are right, it never switches off but it is starved of enough base current to fully conduct.

Brian.
The primaries aren't in series. Its just the way my friend Paul drew it. The 18V sec is just a separate winding, but there is only one Prim winding. This is one transformer. its a big ass 240VA traffo and is toroidal. I did a circuit to switch the taps automatically. Was a pain in the balls to do, but its working great.
 

Hi,
The capacitors alone sorted the issue, but I will add the 100ohm resistors in parallel with the caps, as you have suggested.What wattage resistors should I use?
Also why are the caps solving this issue?

First use the 100R ... then add the capacitors.
The capactiors alone are no stable solution. (It may be now, but connect longer/shorter wires and the problem may be back again!)

Wattage: not important. Small size, no wirewound.

WHY: the capacitors bring a "fast" feedback. The external wires are slow.
--> try to do some slalom course with an RC car. Now try the same but with camera and TV screen, but the TV picture is delayed some seconds. It is much more difficult to keep the car on track.

***
Motor:
Install the Rs first, then test again. It may work, but I´m not sure. An external capacitor may be a solution. But it has it´s drawbacks.
Imagine you try to run a current controlled power LED. You adjust the current to 100mA. As long as the LED is disconnected the current is low and the output voltage will rise to it´s maximum.
Now connect the LED...A huge ammount of pulse current discharges the capacitor until the voltage drops to the LED´s normal operating voltage.
It may be hundreds of Amperes. And for sure this may immediately kill your LED.
--> keep the capacitance at the output low.

Klaus
 

Klaus.
The resistors alone didn't do it. The capacitors alone did. BUT, i will use both as the resistors will help if for some reason the sense wires come loose from my croc clips.
On the output I have a 220uF 63V non polarised cap at the moment (C7). Should I parallel this with a 100nF cap?
 

All of the previous discussion tells me that the circuit's feedback loop is marginal.

Unfortunately, and as others have also noticed, the circuit is quite unusual and overly complicated. It is not readily obvious which of the several resistor/capacitor networks is actually performing the feedback compensation.

There are circuit features that I don't quite comprehend...for instance, what is the purpose of the R14/C5 time constant?

I am going to ask the following question: Has someone else built this circuit, exactly as shown, and made it work? Or is this the very first iteration?

Let me finish with the following statement: I applaud and commend anyone that attempts to build a circuit using only discrete components. Doing so gives you an insight of electronic circuits that an IC does not provide.

But if you are planning to build a discrete transistor supply, why don't you use the "classic" circuit as your starting point? This is a proven circuit that has worked for over 60 years. You may add additional features and enhancements to the basic circuit.

PSU.png
 

It doesn't appear to be based upon sound design principles, rather sort of trail and error approach. I suspect the output will be oscillatory under transient load testing. Maybe the user can enlighten us with a description of how it is supposed to work?
 

I Know a guy who built this exact circuit 35 years ago. His power supply works to this day.
I got the schematic from him. Our layouts are different and I attribute this fact to mine oscillating and his not.
All I know is that the circuit works. I visited him about a month ago and gave it a go. Flawless. Current limiting works great and stable. Better than all 3 of my digital power supplies.
All of the previous discussion tells me that the circuit's feedback loop is marginal.

Unfortunately, and as others have also noticed, the circuit is quite unusual and overly complicated. It is not readily obvious which of the several resistor/capacitor networks is actually performing the feedback compensation.

There are circuit features that I don't quite comprehend...for instance, what is the purpose of the R14/C5 time constant?

I am going to ask the following question: Has someone else built this circuit, exactly as shown, and made it work? Or is this the very first iteration?

Let me finish with the following statement: I applaud and commend anyone that attempts to build a circuit using only discrete components. Doing so gives you an insight of electronic circuits that an IC does not provide.

But if you are planning to build a discrete transistor supply, why don't you use the "classic" circuit as your starting point? This is a proven circuit that has worked for over 60 years. You may add additional features and enhancements to the basic circuit.

View attachment 128138
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top