1
\$\begingroup\$

[UPDATE] I have answered my own question with links to experimental data and theory. Please expand on that answer if you can for the reference of future readers. Thanks for replies so far and ongoing.[/UPDATE]

I have been trying to build a current limiting rectifier on the secondary of a SMPS.

I am building this to be used with any SMPS it is a general purpose concept rather than for the specific use case I am building it for which happens to be a constant current laser driver.

There are many simple driver circuits out there, but they all waste power with a current limiting resistor or with a current sense resistor. The load resistance gets below 1 ohm in my application.

In a simple AC circuit with just a resistive load, you can use capacitive reactance to limit current with very low power loss.

I have been trying to combine this concept with a bridge rectifier to limit the current reaching the diodes to limit the current to the load. This is where the simulation breaks down.

In the screenshot below I have three circuits simulated in LTSpice to demonstrate these ideas.

The first shows the basic example of just a load resistor with current limited by the reactance of the capacitor. I can change the resistance as much as I want below the reactance of the capacitor and the current in the resistor is limited by the capacitor - this is what I want. Of course, a larger resistor would dominate the limiting of current. 12A is my desired peak current as shown.

The second shows the rectifier circuit I will build, which is pretty normal and is there for reference. There is no effective current limiting and the trace shows a peak around 140A.

The third shows my attempt to merge the above two circuits, to limit the current supplied to the diodes to thus limit the current supplied to the load. But the simulation breaks down and tends to a zero current flow.

I have tried tuning the capacitor value.

I have been scratching my head over WHY it breaks down for days.

Thinking about the half cycles and polarity and current pathways etc.

In practical tests I mis-soldered and made a voltage doubler just trying things out. I'm sure there is some theory I am missing. I tried the above with a centre tapped transformer with the same outcomes.

The reason I am doing this is to save power and reduce cooling needs. What actually goes on to tend the current flow to zero and how can I make this circuit work please?

As a secondary set of ideas: I have looked into LM317 and other designs. I even thought about using comparators but the current sense resistor needs to be in series with the load which wastes power and limits current by itself with sub-ohm laser load. I can't find any designs with a current sense resistor in parallel with the load that can have a high 100 ohms to limit power use (compensated for by the gain of an op amp for comparator use.)

Theory would be helpful because I wonder why there seems to be no low wastage method to do this that doesn't need a lot of components such as current transformers.

The voltage sources in the image are all the same 15V AC at 132kHz and represent the secondary coil of a SMPS transformer.

Screenshot of circuits

\$\endgroup\$
7
  • \$\begingroup\$ Usually current limitation in an ac supply is provided by the transformer, as the core saturates, it limits the power to the secondary and rectifier \$\endgroup\$
    – Voltage Spike
    Commented Feb 10, 2021 at 1:06
  • \$\begingroup\$ That's a good point, and something I could depend on if I were building a SMPS from scratch. Or if I were shopping for one, in which case if I couldn't get the exact limit (unlikely) I would have to choose one oversized and then implement a circuit to derate it. In my application, laser diodes experience a current runaway which is why they need constant or at least limited current. They are often modulated so the limiting is more important for me than constancy. Protection/control ICs in SMPS detect the runaway and either shut down completely or modulate at a very low duty cycle, like 1%. \$\endgroup\$ Commented Feb 10, 2021 at 18:14
  • \$\begingroup\$ A better option would be SMPS->Step down DC DC-> Current limiter. If you intend to produce the above circuit in a product, it would not be safe in the event of a fault. \$\endgroup\$
    – Voltage Spike
    Commented Feb 10, 2021 at 20:09
  • \$\begingroup\$ Good thoughts Voltage Spike, I thank you. The above is stepped down and rectified. The current limiter comes after the protection features of the SMPS. I'm not selling this. See me answer below of why it can't work anyway due to the nonlinear nature of diodes. There is experimental data in the links and theory too. \$\endgroup\$ Commented Feb 11, 2021 at 17:46
  • \$\begingroup\$ @VoltageSpike I know this is old, but attempting to pull too much current does not cause a transformer core to saturate. If anything, by causing voltage sag, pulling extra current causes the magnetic flux in the core to decrease. Magnetic flux in core is proportional to volt-seconds, and if frequency is constant, proportional to voltage. \$\endgroup\$ Commented Jan 8 at 18:15

2 Answers 2

1
\$\begingroup\$

I am answering my own question in case someone can benefit. This is a fundamental property of diodes that is causing this. My answer is not robust or complete but it does provide a lot of what you might need to know if you attempt what I did.

These two links have the bulk of the information and I thank "guest" and Edward H Hellen respectively:

capacitor discharge through diode

Verifying the diode–capacitor circuit voltage decay (no paywall)

The short version is the decay curve of a capacitor discharge through a resistor is exponential. However, the decay curve of a capacitor through a diode is much more complicated due to the nonlinear behaviour of diodes. Essentially the decay is logarithmic.

The capacitor is working with AC before it is rectified. The impedance of the diode vastly increases as the discharge voltage from the capacitor approaches zero. This happens on both half cycles. So current only seems to conduct for milliseconds. There seems to be a time delay in the reaction of the diode to this and so the current tends to zero very quickly. Correct me if I'm wrong. Delving into the nonlinear physics that leads to this is something I don't have time for at the moment, because either it is worthy of years of postgrad research or someone has already done it. If you have a reference to this, please link.

\$\endgroup\$
1
  • \$\begingroup\$ Delving into the nonlinear physics that leads to this is something I don't have time for at the moment, because either it is worthy of years of postgrad research or someone has already done it. The good thing is, then, that this deep physics-based understanding is not really needed. Real diodes behave in a certain way, and all you need is some simple electronics bench equipment to acquaint yourself with those. \$\endgroup\$ Commented Jan 8 at 19:52
0
\$\begingroup\$

Since this is AC you can run the conductor through a CT (Current Transformer) without any loss on the primary side. Amperage is not a restriction.

\$\endgroup\$
3
  • \$\begingroup\$ Thanks for that, it's helpful, but I really want to understand why my attempted solution does not work or can't work. I have been thinking about the reactance of the diodes and even if the phase shift of the current makes a difference after the first cap. If someone can point me to the answer that would help I don't mind reading it all up I just need something. As an aside - to actually complete my task I'm making a current mirror which seems to be the simplest solution and works in sims and I have lots of MOSFETs in the junk pile. The set resistor of the mirror runs off 5v with a 0.1 ohm. \$\endgroup\$ Commented Feb 8, 2021 at 19:08
  • \$\begingroup\$ For reference, I built the circuit in Fig 11.7 of wiki.analog.com/university/courses/electronics/text/chapter-11 a great article. I used two 13N03LA MOSFETs with R being judged via LTSpice to be 0.22 ohms with these. Since such a small resistance is temperature dependent, I found that in my actively-cooled circuit I get that much from solder resistance and trace lengths on the board. This is essentially a two-component solution, however other methods must suppress transients down to ps time scale to protect the laser. This is fed from the normal output of the SMPS. \$\endgroup\$ Commented Feb 14, 2021 at 23:38
  • \$\begingroup\$ It's only for short-term testing and not a whole laser driver setup. \$\endgroup\$ Commented Feb 14, 2021 at 23:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.