Is there any variable DC-DC converter suitable for pre-regulation?

[T3sl4co1l]

You may find this helpful for the "computation block":



The op-amp could be the internal error amp of the converter, so you can set variable current limiting, and then the postreg only needs to be a C-mult.

Tim

[bloguetronica]

You may find this helpful for the "computation block":



The op-amp could be the internal error amp of the converter, so you can set variable current limiting, and then the postreg only needs to be a C-mult.

Tim

I'm not sure if that was what I had in mind. How does that circuit work? I see some kind of current mirror to limit the current (transistor at the bottom), then the input stage of an op-amp, and then a transistor pulling up the negative input of the amp-op, which I presume is to implement CC limiting.

Kind regards, Samuel Lourenço

[T3sl4co1l]

Right, the transistor stuff wraps around the existing error amp, say of a CV power supply to add CC operation.  The transistors have limited gain, a necessity to deal with the already high loop gain of the op-amp without introducing too much slop themselves (it will transition from CC to CV mode in the span of maybe 100mV at the output pin).

This is better than two opamps wired-OR with diodes, because there is no integrator windup.  It acts ~instantly, going from CC to CV mode.

Tim

[Wolfgang]

Hi, a design I saw was from Jim Williams with an LT1074 preregulating an LT1083. ChangPuak also used this idea in a wide range lab PSU:

https://www.changpuak.ch/electronics/PETH-581.php

[bloguetronica]

Hi Wolfgang,

I just don't want to copy a design. I want to build a design that I can understand. That is one of the points of the project. Anytwy, I appreciate your suggestion.

Right, the transistor stuff wraps around the existing error amp, say of a CV power supply to add CC operation.  The transistors have limited gain, a necessity to deal with the already high loop gain of the op-amp without introducing too much slop themselves (it will transition from CC to CV mode in the span of maybe 100mV at the output pin).

This is better than two opamps wired-OR with diodes, because there is no integrator windup.  It acts ~instantly, going from CC to CV mode.

Tim

Thanks Tim!

Anyway, I've decided to ditch the whole computational block idea, since it wont calculate the voltage well. So, I will base the DC-DC pre-regulator feedback voltage on the output voltage via the PNP transistor. I decided to implement constant current on another way, using a transistor to starve the base of the pass transistor (perhaps using a MOSFET instead of a BJT to starve the base of the main transistor is better?).

That transistor (BJT or MOSFET?), in turn, will have its base connected to the output of a op-amp wired as a comparator. The reason I'm using an op-amp here instead of a comparator is because I want the op-amp to operate in its linear region. At the inputs of the op-amp, is connected the output of an INA180 current sensing amplifier and the output of a DAC that sets the current.

I've also decided to ditch the voltage sensing. After all, I don't need the extra precision. I'll post the drawn schematic later.

Kind regards, Samuel Lourenço

[Wolfgang]

Hi,

I never suggested to build something you dont understand. I never do that, either.
Nobody stops you from learning and understanding designs from others (including simulation) and improving or adapting them on your own.

[bloguetronica]

Sorry for misunderstanding. I never taken you as being that "assertive", and it was not my intention to do so, hence the word "suggestion".

Anyway, I have an idea for a schematic, but I'm having a busy day. Will post it as soon as possible.

Kind regards, Samuel Lourenço

[T3sl4co1l]

So you want something you created and understand, but that is worse than the standard approach?

Tim

[bloguetronica]

Hi Tim,

How it is worse? I didn't presented the schematic yet!  What is the standard approach? Am I taking the wrong approach by basically not copying a design (at least blindly)? Although Wolfgang presented a pretty much valid and working design, I decided not to use it, but thanked anyway.

What can I say? I like to reinvent the wheel, despite the fact of often being criticized for that. I like to experiment, and even if I do mistakes it is not important, because I can learn from them. Nothing beats our own experience.

I also don't know the concept of "standard" in electronics. There is just different ways of doing things. And since I was over-complicating I just decided to be back to the drawing board (the "computation block" would never work as intended, BTW, because I was thinking wrong in the first place). Simpler, more elegant design, with less problems and caveats.

Regarding the pre-regulator DC-DC, I'll use the PNP transistor to control it, and it will be based on the output voltage. As you suggested earlier:

You don't want the SMPS output to track DAC/setpoint output, because it won't follow under a current limit condition.  That's why it's made to follow the actual output instead.
...

Kind regards, Samuel Lourenço

[T3sl4co1l]

It's a behavior I've seen a lot in audiophile circles, and that's probably why I snapped at it.  They can get away with it, because audio amplification is a very easy problem (tolerant of poor solutions), and few of their designs ever have to face the rigors of reality.

It works, for some range of "works", but that doesn't mean it's a good idea.

It's very much a good thing to be different, to explore the available design space -- there are many valid ways to wire up components (although very, very many more invalid ways in total!), all with different tradeoffs.  Having the experience to know which combinations of tradeoffs are best for a given application, is the payoff.

The problem is if you get far enough into the project that it works, then you stop without going the extra step to finish it up, to make it work consistently and reliably -- that's the part that bugs me the most.

Another way to put it: is this a learning experience, or a challenge towards a stated goal?

If it's the former, definitely do play around and test -- emphasizing in particular, the many ways a circuit can go wrong.  Test the extremes of inputs and outputs, of environmental conditions (temperature and such), of component ratings.  See how things fail, and understand why manufacturers choose ratings as they do.

While, if it's the latter, you have a clear goal in mind -- a power supply with some minimum efficiency, maximum noise, operating range, functionality and so on (probably with size and cost being secondary considerations) -- there is a far smaller space to choose from, that will meet those goals.

I'm probably wrong for emphasizing a goal here, and that's probably because my experience points me, very quickly, to a set of possible solutions.  I should reflect more often on what it was like without that experience...

Tim

[bloguetronica]

I think you are mixing the fact that I want a linear power supply with what happens inside the audiophoolery circles. These are two separate goals. One involves (me) designing a power supply with acceptable noise levels, even if it is not 100% efficient, the other revolves around expensive capacitors, unidirectional cables and snake oil.

There are far lesser efficient PSUs than the one I'm designing now, believe me (and that includes two that I've designed myself, that are very inefficient, emphasis on very). The DC-DC pre-regulator is not the center of the design, but a secondary aspect that is there just to avoid excessive dissipation on the pass element. Making it the primary regulation element, and then adding a pretty heavy filter, may not solve the noise completely. In fact, I consider using a DC-DC converter a necessary evil, just to keep the design more compact. I'm just using it, because the evil is already done since the whole contraption will be fed by an isolated and unregulated DC-DC converter.

This is just a concept that I'm developing, that may turn or turn not into a fully fledged project. And, if that fails, a learning experience. Essentially, most of my projects started that way.

Kind regards, Samuel Lourenço

[NiHaoMike]

I think you are mixing the fact that I want a linear power supply with what happens inside the audiophoolery circles. These are two separate goals. One involves (me) designing a power supply with acceptable noise levels, even if it is not 100% efficient, the other revolves around expensive capacitors, unidirectional cables and snake oil.

There actually are true directional cables - they have multiple layers of shielding with the outer layers grounded at only one end. The intent being that induced noise gets conducted to the device that is less sensitive to noise (usually the source) or has a more direct path to ground.

[bloguetronica]

I think you are mixing the fact that I want a linear power supply with what happens inside the audiophoolery circles. These are two separate goals. One involves (me) designing a power supply with acceptable noise levels, even if it is not 100% efficient, the other revolves around expensive capacitors, unidirectional cables and snake oil.

There actually are true directional cables - they have multiple layers of shielding with the outer layers grounded at only one end. The intent being that induced noise gets conducted to the device that is less sensitive to noise (usually the source) or has a more direct path to ground.

I think you should elaborate on that. If shielding is well done, the behavior of the cable should be the same lengthwise. But we are drifting a tad off-topic here.

Kind regards, Samuel Lourenço

[T3sl4co1l]

I think you are mixing the fact that I want a linear power supply with what happens inside the audiophoolery circles. These are two separate goals. One involves (me) designing a power supply with acceptable noise levels, even if it is not 100% efficient, the other revolves around expensive capacitors, unidirectional cables and snake oil.

FYI, there is a more technical, design-centered facet of audiophoolery, that likes to play with novel circuits -- diyAudio and Audiokarma are probably good examples.  Audiophoolery is a huge sliding scale, from good engineers with a bit of a kink (an analogous and somewhat topical example might be Dave Berning's DC-transformer (chopper based) designs, worth a look), to the casual rich, looking to improve the chakra of their music with crystal pyramids supporting hand-woven silver cables.

Don't take it as an insult -- it's a sign of creativity.  Just don't get disgusting with it, that's all.

Making it the primary regulation element, and then adding a pretty heavy filter, may not solve the noise completely.

Let's talk about that --

How much attenuation is "complete"?  40dB (100x)?  80dB?  200dB?  Below the noise floor (whatever that might happen to be)?

Because it of course can never be complete-as-in-actually-zero, not in the real world, not at finite temperature and over finite time!

How much attenuation do you expect to obtain from the linear regulator stage?

It's not as easy a problem as you may envision, and you're open to common mode noise regardless of what circuitry is on either side of the converter.  The converter needs to be laid out properly to avoid this, regardless, and some additional filtering will inevitably be required (for some values of "complete" ).

So anywhere you're using a nasty rattling converter, you have to be careful to filter it, on all sides.  The only true solution is going full linear, which as you already understand is quite inefficient, and is still not the greatest because you still have to deal with mains noise, diode recovery noise, and supply ripple.  The noise floor is set by the output transistor, and/or overall output impedance*, which defines how much filtering you need, to reach that floor from some given noise source, whatever that might be.

*Note that capacitors have noise, just as everything else does: take the impedance at a given frequency, and convert it into noise with the Johnson noise relation:
https://en.wikipedia.org/wiki/Johnson%E2%80%93Nyquist_noise#Thermal_noise_on_capacitors
Since the impedance is usually low (~ohms), this is a small contribution (below 1nV/rtHz), but still nonzero.

The easiest low-noise source is a battery: the effective capacitance is extremely large, and as long as the current draw is small, the chemical overpotential is small as well.  Noise is usually due to thermal and ionic motion, having a 1/f spectrum; keeping wind currents away, and current draw low, helps greatly.

Anything else is more complicated, whether a linear, switching or hybrid supply.

I recommend a well-filtered switching supply, because you may need filtering regardless of what topology you choose, and filters are easier to make than low-noise linear regulators: just throw more inductors and capacitors at it!

Tim

[bloguetronica]

Hi,

Didn't take any offense, not at all. However, I was thinking that you took me for an audio-follower or something.

As for the filter, I'm not concerned about its attenuation, but what can be achieved with varying loads. Using a filter right at the output is kind of novel, although a DC-DC already uses an inductor and a capacitor, basically a filter. Going full linear unfortunately is not an option, for the reasons mentioned above. So I think it is better to go midway and compromise.

Anyways, using a low  to medium frequency DC-DC converter is within my plans. Although lower frequencies are harder to filter, typically the PSRR of the linear stage is much better for those frequencies.

Kind regards, Samuel Lourenço

[bloguetronica]

Hi,

I've been very busy lately. Only today I've came up with a schematic. The DC-DC pre-regulator will be designed according to the "PNP tracker" configuration shown.

There are some things that I want to explain about the schematic:
- The first half of the OPA2703 is required for buffering, while the second half is responsible for the feedback control. The added resistors will provide a gain of four over the voltage from the AD5551 DAC, while also providing voltage sensing and lead compensation. The OPA2703 was chosen because of its low offset. This block will require 0.05% resistors to maintain the overall precision.
- The current control is done via the LTC2640 DAC. The INA180 is an instrumentation amplifier specialized for current sensing, and the OPA705 works as a comparator of sorts. I didn't used an actual comparator here because I want the OPA705 to operate in its linear region. The 2N3904 will starve the base of the pass transistor.
- The pass transistor is a MJD122 Darlington. Unfortunately, it is not properly drawn.
- The 0.1 resistor will make sure that the inverting input of the INA180A will not be pulled down to 0, maintaining the operation in the linear region.
- I'll have to replace the OPA705 and OPA2703 with higher voltage parts, or decrease the supply voltage and also the voltage gain (overlooked that part). Edit: I will use the OPA2180 as a replacement for the OPA2703. The OPA705 is fine to use, since I can supply 5V to it.

Kind regards, Samuel Lourenço

[bloguetronica]

Hi,

Been simulating the voltage regulator circuit with current limiting, and it worked as expected. VG1 is the reference that sets the output voltage, which is 4x bigger than the value of this reference voltage. On the other hand, VG2 sets the current limit to a value that corresponds to this reverence voltage divided by four. In other words:

Vout = 4 x VG1
Iout = VG2 / 4

I've used the OPA703 in the simulation, instead of the OPA705 I'm going to use in the end product. The transistor at its output can be either a BJT or a MOSFET (the later should have a low Vgs). In the near future, I'll implement the DC-DC pre-regulator circuit in the simulation. So far, it worked like a charm, even when defining small output voltages and currents.

Attached is a DC-DC sweep of the reference voltage VG1. You can see how the output voltage increases until the current limit is reached, because a 16 load resistor was used, and the current limit is set to 500mA (VG2 = 2V).

Kind regards, Samuel Lourenço

[bloguetronica]

Hi,

Meanwhile I did a few tweaks to the circuit. I've added a CC LED indicator that light up when the threshold current is exceeded. Basically, the output of the OPA703 goes from approx. 0V to around 600mV, which is the voltage required to bias the CC control transistor. The comparator compares that to a reference voltage of around 300mV, given by the BAT54, and lights up the LED when the CC condition is detected.

I've noticed that the base current of the transistor is very stable, and stays around 60 to 70uA when the circuit is in CC mode. So, there is no need of a base transistor to limit the current, since, as expected, the negative feedback loop takes care of that. Thus, the transistor is always correctly polarized and works in the active region when passing current.

Kind regards, Samuel Lourenço

[T3sl4co1l]

An op-amp driving a base directly?  Oh dear.

Tim

[bloguetronica]

An op-amp driving a base directly?  Oh dear.

Tim

Hi Tim,

What will happen? I'm trusting the simulation too much? A base resistor doesn't seem to make a difference there, since the current is too small. Do you think it might over-saturate the transistor in case of a transient, case I don't use a resistor? Or is is just a convention?

Kind regards, Samuel Lourenço

[T3sl4co1l]

Yeah, you completely made an oscillator, U3 has no feedback or compensation and the transistor is wide open.  It might simulate stable under some circumstances but it will definitely oscillate in practice.

Tim

[bloguetronica]

Hi Tim,

Did some transients and steady state simulations, and indeed that oscillation happens. So, I've decided to stop using an op-amp as a comparator and decided to use the LMV331 comparator instead. The datasheet doesn't mention an hysteresis figure, and also suggests that can the comparator can be used where no hysteresis is desirable (with the due PCB layout cautions).

So, I've decided to redraw the circuit. But the transient and steady state simulations seem to show convergence errors. The DC sweep works as expected. The new circuit shows a much more precise current limiting (indicative of problems with the old circuit). It seems I'll have to build the circuit using the comparator to see how well it works. The pre-regulator will be figured out later, but I'll use the bipolar transistor solution presented before here, somewhere. The CC LED indicator will not be used, because it is not possible to use in this solution without messing up the circuit.

See "Sim2.png" for the new circuit. A comparator should be stable. I didn't used before because I was worried about the hysteresis that these guys normally show. However, since I have some degree of negative feedback, the voltage at the inputs will have a tendency to be equal in the CC condition (as also simulated with the OPA703 op-amp). That means the comparator will work in its "active" region (perhaps "unstable" by other means).

Thanks, and lets see how this goes.

Kind regards, Samuel Lourenço

[T3sl4co1l]

Nooooo no no no!

A comparator just oscillates more!

Are you trying to make a fault latch instead?  Then you need a latch.  You can't have a comparator turn off the signal that turns it on, that's the definition of a (relaxation) oscillator.

But a stable, linear current limit, just control the loop gain and phase so it's stable.  Which means, limiting the gain of that transistor (typically, a base voltage divider and an emitter degeneration resistor are added), and compensating the amp (an R+C from OUT to -IN, and a series resistance to -IN from whatever source is feeding it, if necessary).

Edit: also, have you checked the LMV331 datasheet?  It's a 5V part, and propagation delay is probably as high as the rest of the circuit.  Sanity check!  I don't know what model is inside that part there, but I suspect if you test it alone you will find it bears no resemblance at all to a real LMV331!

Tim

[bloguetronica]

Hi Tim,

No, a fault latch is very undesirable. I just want to implement current limiting, and that is why I opted for an op-amp in the first place, because I want the circuit to operate in the linear region (no hysteresis whatsoever, in any shape or form). I was expecting that this comparator could be used like so, since it is stable in open loop and supposedly has no hysteresis. If it has, if even by a tiny ammount, then the whole circuit would act indeed as a relaxation oscillator.

So, how do I limit the gain of the transistor? How do I stabilize the unstable op-amp? R+C in parallel from the output to the input? Or in series? A rough schematic of this sub-circuit will do. Then I would tweak the values. I can't find examples on how to use an op-amp as a comparator of sorts.

As for the 5V supply, it is implemented in the simulation and will be implemented in real life, from a completely independent regulator that is fed by the 24V rail. No worries!

Kind regards, Samuel Lourenço

[T3sl4co1l]

Your problem is thinking it's a comparator at all -- a comparator is a one-bit analog-to-digital converter.  An ideal comparator has no linear range whatsoever.  We could maybe use one in a PWM circuit, but there's nothing linear and stable about one!

A fault latch isn't uncommon in power control functions -- if not in bench supplies -- and that would be an expected use for a comparator.  A latch is a digital logic element, so a comparator would trigger it -- hence why I asked to make sure.

Rather, what you're looking for is more of a "fuzzy logic" condition: you want the current error amp to be inactive (saturated output-low) when the output current is in the range 0...IREF, and active above there.  What's more, you require active operation to be limited in speed, so it isn't chasing its tail (oscillating) in the active range.

Yes, R+C in parallel with the amp, you may remember this:
https://www.seventransistorlabs.com/Images/Limiter2.png
The op-amp is the current error amp, and the R+C across it is the compensation.  R5 is the series input resistor.  In this case, you can ignore everything else in the circuit (but using this circuit instead of the two error amps is far superior!).

Tim