Interruptible Switch Mode Power Supply?

Thread Starter

Shanjaq

Joined Dec 6, 2010
12
I'm trying to figure out how a Switch Mode constant-current Power Supply could work with a TTL-controlled laser module on its output. If I'm not mistaken, the constant-current is maintained by an Inductor in series with the load. My concern is that when the load is disconnected, the Inductor will instantly produce a high-voltage spike because it suddenly sees an increase in load resistance to infinity and all that "electrical momentum" basically slams into a wall.

I've thought maybe it would work to shunt the current away from the load using a sequence of switches (open alternate path, close main path) but that could be problematic as it has to happen in a consistent order and timeframe otherwise the Inductor will destroy the switches with an HV transient.

Another thing I thought about briefly was to just disable the oscillator, but the Inductor would simply coast down over a (relatively) long period of time and the laser would not shut off instantly, which is what I need.

Are there any well-known and reliable methods for rapidly toggling the constant-current through a Switched-Mode Power Supply's load?
 
Last edited:

AnalogKid

Joined Aug 1, 2013
10,987
If I'm not mistaken, the constant-current is maintained by an Inductor in series with the load.
You are mistaken. An output inductor in a switching power supply can be doing several different things depending on the circuit topology, but maintaining a constant output current is not any of them. A constant current (CC) circuit usually involves a resistive current shunt and a control amplifier. In the case of a true CC output switcher, the switch driver pulse width is modulated based on the average output current.

ak
 

Thread Starter

Shanjaq

Joined Dec 6, 2010
12
Here's a picture of my general setup:


If there were a switch on "Vout", when opened it would cause C1 to charge off the inductor L1, which would cause a direct short of all that reactive power stored in the cap through the load on the next positive cycle of the PWM (which would be a lot of current for an LED!)

How do you rapidly toggle current through a load (whatever's connected to Vout,) with a setup like this?
 
Last edited:

AnalogKid

Joined Aug 1, 2013
10,987
Most of your description is incorrect. If a load at C1 is switched off rapidly, then any inductive kick from L is absorbed by C1. After all, that is exactly what happens every switching cycle in a boost converter - the high voltage inductive kick is where the "boost" comes from.

The second half of your explanation doesn't make much sense.

Except for the average current control modulation, a PFC boost converter is no different than any other flyback boost converter, although it will have greater than normal ripple across C1. Note that Q1 puts "a direct short" on the output of the inductor every cycle. This is what charges up the inductor.

ak
 

Thread Starter

Shanjaq

Joined Dec 6, 2010
12
You're not looking at what's not there; Imagine there's a switch "Q2" and an LED to ground hanging off the regulated output "Vout".

So let's say that "C1" is 10uF and the inductor "L1" is 10mH while the supply "Vin" is 12v. As long as Q2 is closed, the current-regulating process will continue to operate normally, let's say the PFC is configured to deliver 120mA constant current. As soon as Q2 opens the output circuit, C1 will begin charging off the flyback for each PWM cycle. Leave Q2 open for a couple seconds, and C1 will reach a fairly high voltage from all that flyback energy, several hundred perhaps.. Now close Q2. What happens to the LED?
 

AnalogKid

Joined Aug 1, 2013
10,987
You're not looking at what's not there; Imagine there's a switch "Q2" and an LED to ground hanging off the regulated output "Vout".
Imagine that this is an electrical engineering forum, and a schematic is posted without any explanation of the secret invisible parts.
Imagine the schematic is for a high power AC circuit, but the application is for a low power DC circuit.
Imagine the quality of the responses.

Please post a schematic of the circuit you wish to discuss. To paraphrase Admiral Josh Painter to Jack Ryan, engineers don't take a dump without a schematic.

ak
 

kubeek

Joined Sep 20, 2005
5,794
...As soon as Q2 opens the output circuit, C1 will begin charging off the flyback for each PWM cycle...
What you are missing here is the feedback of the PWM controller. It senses the current flowing through the Vout terminal and as soon as you disconnect the load it decreases the duty cycle to zero or whatever is needed to maintain the set current.
So in other words, nothing like you describe will happen unless the PWM is static and the whole circuit runs without any feedback.
 

AnalogKid

Joined Aug 1, 2013
10,987
So let's say that "C1" is 10uF and the inductor "L1" is 10mH while the supply "Vin" is 12v. As long as Q2 is closed, the current-regulating process will continue to operate normally, let's say the PFC is configured to deliver 120mA constant current. As soon as Q2 opens the output circuit, C1 will begin charging off the flyback for each PWM cycle. Leave Q2 open for a couple seconds, and C1 will reach a fairly high voltage from all that flyback energy, several hundred perhaps.. Now close Q2. What happens to the LED?
Are you really asking about applying power factor correction to 12 VAC? (Note that simply stating Vin as 12 V is dangerously ambiguous.).

At its heart a boost converter is a voltage-mode output circuit. It can be configured as a current-mode output by placing a current shunt sense resistor in series with the output and adding signal conditioning circuits and somehow tying this into the pfc controller, but that is a significantly different circuit. Also, to maintain constant current there must be feedback from the output to the "pfc" controller. This feedback will keep the converter output from exceeding its design parameters, unless it is badly designed.

Also, A boost converter is not a charge pump, so the output will not reach a fairly high voltage if unloaded.

Re-reading your post, it sounds like there is no feedback from the load to the controller, and you are relying on the load impedance and the converter circuit's output impedance to form a voltage divider, and relying on the circuit's output impedance to limit current into the load. While this approach may work for one specific set of conditions and components, it is not a stable design approach.

ak
 

Thread Starter

Shanjaq

Joined Dec 6, 2010
12
It can be configured as a current-mode output by placing a current shunt sense resistor in series with the output and adding signal conditioning circuits and somehow tying this into the pfc controller, but that is a significantly different circuit. Also, to maintain constant current there must be feedback from the output to the "pfc" controller.
^
This quote is probably the closest to what I'm doing. So assuming all of the above is in place, what happens when the load resistance increases? Shouldn't the CC control loop attempt to compensate for the decreased current by boosting the output to overcome the additional impedance? What, then, will happen when the impedance reaches infinity in a very short time period? How is that not going to result in maximum boosting as the circuit attempts to force a set current through a high impedance?
 

AnalogKid

Joined Aug 1, 2013
10,987
This quote is probably the closest to what I'm doing.
"probably" the "closest" ?!?!? - what a ridiculous way to describe an electronic circuit.

This has gone on long enough. After 9 posts and still no schematic, it is clear that you are not interested in an answer

ak
 

Thread Starter

Shanjaq

Joined Dec 6, 2010
12
Anyone else care to take an objective look at the requirements and comment on the concerns I've tried to illustrate? What are some good CC supply designs which support having their load switched in and out at high frequencies?
 

RamaD

Joined Dec 4, 2009
328
I think the answers are all there in the posts. A CC supply will have a compliant voltage, meaning that if there is no load, the output voltage will be the compliant voltage. The response will be decided by the loop. When you turn on, it would take some time to reach the final CC value.
Instantaneous load removal, with a series inductor, will produce a kick which will be absorbed by the capacitor.
Your requirement appears to be a fast responding CC supply, is that right? If you quantify your requirements, you will certainly get answers here, like I have always got.
 

kubeek

Joined Sep 20, 2005
5,794
It really depends on the load. Assuming the load is ok with having the current abruptly go to zero, then you could try turning the load off and turning some other load with similar effective resistance on. That should achieve the least change in the output voltage and feedback loop.

Couldn´t you just post what the actual goal is? It allways is much easier to help you with the real problem than with your solution.
 

Bernard

Joined Aug 7, 2008
5,784
It isn't easy to find a schematic for a potted CC LED driver. Post by White G, Light & Sound Sculpture, has the same problem with dimming
50W LED from non dimming CC supply. My thought , post # 23, was to dump PWM off part of cycle into a dummy load. I do not have a SM CC supply to play with & my Nam era oscilloscope died. Wonder if load dumping is really necessary??
Seems that the goal is to pulse a laser.
 
Last edited:
Top