ZXLD1370 Buckboost LED driver can be that accurate?

Hello,
The ZXLD1370 buckboost led driver offers very accurate led current regulation even though it puts the sense resistor in series with the inductor and not the LEDs.

Do you believe that the graph of figure 18 on page 13 of the datasheet would show the led current not varying much with input voltages extending up to 30V?

Achieving the current accuracy claimed in the datasheet (in spite of line and load voltage changes) in buckboost mode when the led current is not being directly sensed, is, I think you will agree, pretty fantastic. Can it really be this good? No other IC datasheet makes these claims.

ZXLD1370 datasheet
**broken link removed**

Also, why in the zxld1370 datasheet, on pages 2, 15 and 36, does it show the output capacitor going to ground, instead of being across the leds, when in buckboost mode?

Hysteretic regulators with 1% performance have been around for many years, but Zetex must have used their expertise in very low saturating switches and very fast recovery times to achieve these performance features. I like the thermistor regulation for 75'C derating the output drive.

The nearby output Cap of 4.7uF is just there to smoothen the LED current edges and reduce EMI well above 1 MHz.

With LED ESR's in the 1 ohm range, it suppresses large loop currents rise times with 5us time constant. Smaller values could be used with long strings.

Connecting the cap across the LED's would inject noise back into the supply source.

This chip has my vote.

I agree , it has my vote too, I have never seen a led driver in buckboost that doesn't sense the led current directly and yet still regulates it accurately as described by the zxld1370 datasheet.
I wonder if they only show input voltage regulation up to 17v because above that, the led current regulation deteriorates?

Dont confuse the accuracy of Current control with Rs which is a necessary and the 1st diagram which show thermal feedback used to drive a string of LEDs with soft start up to maximum power until it reaches 75'C and then reduces power to zero at 85'C, so it regulates between there depending on thermal resistance of connection to one LED or heatsink.

In this application Rs=0, so current adapts to LED current, and power dissipation capability of LEDs and heatsink thermal resistance.

Ref: Buck-Boost Diagram Utilizing Thermistor and TADJ

ZXLD1370 appears to be monitoring the peak and trough current levels of the inductor. As you know, the inductor current doesn't always flow through the leds, -when the fet is on the inductor current chrarges up through the fet and yet zxld1370 still "knows" what is the led current.
So basically Rs isn't always in the loop of current with the leds, and yet still they are able to regulate the led current via Rs...but how well, does it work with vin changing from 10 to 30v?

I don't see how their controller compensates for duty cycle variation. Equation 3 seems to be pulled out of thin air, where does that 1/(1-D) part come from? Do they internally measure D and use it to vary Vsense or something?

Also I fail to see the point of measuring LED current indirectly in the first place...

Equation 3 seems to be pulled out of thin air

Click to expand...

I am so relieved to find someone else also thinking that......I have been staring at it and wondering from where it came....but the explanation there (just above equn 3) appears to say that the controller does indeed measure the duty cycle...and yes the zxld1370 does appear to adjust vsense in relation to duty cycle.

The point of measuring the current like they do is that you can do it with hysteretic control which means no loop compensation needed....also, you get much less overshoot when flashing on and off, and also better response for pwm dimming of the leds. (greater dimming range)

I am now trying to also work out how the zxld1370 regulates the switching frequency to about 400KHz. (it says it does this) Surely, if the zxld1370 is monitoring the vsense voltage peak and troughs, then the inductor value determines the switching frequency, along with the input and output voltages?

treez said:

The point of measuring the current like they do is that you can do it with hysteretic control which means no loop compensation needed....also, you get much less overshoot when flashing on and off, and also better response for pwm dimming of the leds. (greater dimming range)

Click to expand...

You will get nice rise and fall on the converter's input current, but not on the output current. But the mystery of the self-adjusting Vsense also makes me doubt any assumptions about its dynamics.

I am now trying to also work out how the zxld1370 regulates the switching frequency to about 400KHz. (it says it does this) Surely, if the zxld1370 is monitoring the vsense voltage peak and troughs, then the inductor value determines the switching frequency, along with the input and output voltages?

Click to expand...

Right, the datasheet presents many contradictions, and apparently several typos as well... generally I would avoid parts like this, unless they make a demo board that allows it behavior to be easily observed.

There is a demo board, ZXLD1370EV2, but it does not seem to be available anywhere...

**broken link removed**

You will get nice rise and fall on the converter's input current, but not on the output current

Click to expand...

....it surely will be the same for output current, I mean, the ZXLD1370 should operate wthout an output capacitor, as its hysteretic, and the sense resistor is meant to sense the full ripple in the inductor current , as it senses the peaks and troughs of the inductor current......thus surely it will be pretty devoid of output current overshoot, unlike what one might see using an alternative part such as LM3421...buckboost schematic on page 56 of LM3421 datasheet

https://www.ti.com/lit/ds/symlink/lm3421.pdf

I suppose it would be nice to just avoid it, but if the ZXLD1370 really does accurately regulate led current in spite of either vin or vout changes, then it will be worth its weight in gold

Given the expertise of Zetex and their patents in this design, I would not expect them to give away their secrets for dynamic power dissipation control and precise regulation.

One must use their formula and if they don't work, then there is cause for concern. How they achieve this performance will be their secret unless you can read between the lines in their patent applications.

However the optimal use of Boost converters always uses a combination of PFM and PWM to obtain a wide dynamic range of load control and input regulation is not a new practise.

You could contact Diodes Inc directly for availability of ZXLD1370EV2, since their distributor has no part listed.

This is the response I just had from diodes.com

Thank you for contacting Diodes.



I have been passed your inquiry from Inquiries Europe.



Equation 4 shows that the LED current is independent of

input voltage and output voltage.



Equation 3 shows the Sense Resistor current IRS.

ILED is IRS*(1-D) so the (1-D) term in IRS is cancelled out

leaving ILED independent of D hence VIN and VOUT.



If VADJ=VREF then Equation 4 simplifies to

ILED=(VSENSE/RS)*GI_ADJ.

Click to expand...

...they are just basically saying what it says in the datasheet, I am hoping someone here will know whats going in behind this?

The circuit operation principle, which is discussed as gating function in the datasheet, is explained more detailed in the Zetex Diodes patent application, e.g. EP2234255. I agree with Zetex support people that the duty cycle cancellation as such is plausible.

Of course understanding the principle doesn't involve knowing the practical issues or e.g. calculating the accuracy. This is more a matter of empirical measurements, I suppose.

FvM, I hope you are right, because I am sure you will agree that if this hysteretic type buckboost chip can really regulate ILED over significant load and input line changes, then we are looking at an absolute utter revolution in LED driving here.

treez said:

....it surely will be the same for output current, I mean, the ZXLD1370 should operate wthout an output capacitor, as its hysteretic, and the sense resistor is meant to sense the full ripple in the inductor current , as it senses the peaks and troughs of the inductor current......

Click to expand...

The input and output current are related by a RHPZ though. For example if you apply a negative step to the current setpoint and duty cycle, you will get overshoot on the output current. For example, if your converter is steadily operating at D=0.75, and you suddenly drop it to D=0, your input current will indeed slope down gracefully, but your output current will instantly quadruple before decaying down to zero. Hysteretic control doesn't make this phenomenon magically disappear. In practice, there is always a tradeoff between control bandwidth and overshoot.

Surely Hysteretic converters don't have RHPZ's?
FvM's patent article tells that the ZXLD1370 regulates the peaks and troughs of the inductor current, such that it changes D to allow the ILED to stay the same when either vin or vout changes....as the datasheet etc says, the inductor current can never go above the level of [(average inductor current) + 15%]...so you can't get big overshoot , surely?

treez said:

Surely Hysteretic converters don't have RHPZ's?

Click to expand...

The converter plant itself, when configured as a boost or buckboost, will have a RHPZ in the duty cycle to output current transfer function. That will never change.

FvM's patent article tells that the ZXLD1370 regulates the peaks and troughs of the inductor current, such that it changes D to allow the ILED to stay the same when either vin or vout changes....as the datasheet etc says, the inductor current can never go above the level of [(average inductor current) + 15%]...so you can't get big overshoot , surely?

Click to expand...

You would not get overshoot on the inductor current, since that is the current on which the hysteresis acts. But the dynamical relationship between the inductor current and LED/output current does not change. It's still the same old transfer function, RHPZ and all. I invite you to simulate it if you're skeptical.

sorry but I cannot simulate it because diodes.com don't give enough away about the "gating function" which they say does the control.
As you know, being hysteretic the ZXLD1370 based bukboost can have no output cap......in such case the peak led current can never go above the peak level of the inductor current.

Regarding the compensation cap for the zxld1370, its just recommended to be 100p in all cases...so its obviously just for noise purposes...so realistically, there is no compensation network with an zxld1370 based buckboost, and surely a bukboost with an rhpz could never work without a significant compensation network?

treez said:

sorry but I cannot simulate it because diodes.com don't give enough away about the "gating function" which they say does the control.

Click to expand...

The gating function thing won't change whether the RHPZ is there or not (unless it happens to have a DSP built in). That function is just there to address steady state behavior, not dynamical behavior.

As you know, being hysteretic the ZXLD1370 based bukboost can have no output cap......in such case the peak led current can never go above the peak level of the inductor current.

Click to expand...

Of course there should be an output capacitor, every schematic in the datasheet even shows one. And they show it connected to ground, which seems like a terrible idea in the buckboost configuration... better to be put directly in parallel with the LEDs. And when we talk about overshoot, naturally we're referring to the average current per switching cycle, not instantaneous peak current.

Regarding the compensation cap for the zxld1370, its just recommended to be 100p in all cases...so its obviously just for noise purposes...so realistically, there is no compensation network with an zxld1370 based buckboost, and surely a bukboost with an rhpz could never work without a significant compensation network?

Click to expand...

I'm not even going to attempt to explain how their compensation works, given that the documentation simply doesn't give adequate information. But I don't think the controller will be unstable. I'm saying that the output current will have overshoot on it if you use the PWM dimming function in a boost or buckboost configuration.

Of course there should be an output capacitor

Click to expand...

The zxld1370 in buck mode definitely doesn't need an output capacitor, so why should the zxld1370 in buckboost mode need an output capacitor?......either way its just controlling the peak and trough points of the inductor current....and by the way, I think it does have a dsp (or similar) inside it.

Regarding the output capacitor, page 16 says (under "application circuit design") that the output capacitor "if needed".....bla bla....so this proves that in bukboost mode the output cap is optional...otherwise, why would it say "output capacitor (if needed)"

**broken link removed**

I once did a ~5W sepic converter led driver running of a battery with a different controller than zxld1370....it was a hv9910b with an external opamp error amplifier...it actually regulated its own input current (the battery discharge current)......it actually worked well and switched smoothly with no output capacitance at all....(other than 100nF or so which was needed to stop overvoltage ringing at the output).....so ifthat can have no output cap, then im sure the zxld1370 bukboost doesn't need one either...because the sepic has an rhpz just like the bukboost.

If the zxld1370'ssense resistor was always in the led current loop, then yes I would agree that it would suffer the rhpz...but it is not, it is in the same place as where it is when the zxld1370 is a buck converter...the zxld1370 doesn't know if its connected up as a buck or a bukboost, surely?

treez said:

The zxld1370 in buck mode definitely doesn't need an output capacitor, so why should the zxld1370 in buckboost mode need an output capacitor?......either way its just controlling the peak and trough points of the inductor current....and by the way, I think it does have a dsp (or similar) inside it.

Regarding the output capacitor, page 16 says (under "application circuit design") that the output capacitor "if needed".....bla bla....so this proves that in bukboost mode the output cap is optional...otherwise, why would it say "output capacitor (if needed)"

Click to expand...

Because sometimes datasheets say dumb and wrong things? LEDs don't like high peak:average current ratios.

If the zxld1370'ssense resistor was always in the led current loop, then yes I would agree that it would suffer the rhpz...but it is not, it is in the same place as where it is when the zxld1370 is a buck converter...the zxld1370 doesn't know if its connected up as a buck or a bukboost, surely?

Click to expand...

The RHPZ isn't a function of the controller. There is literally nothing the controller can do to eliminate the RHPZ (except run the thing in DCM) in a CCM boost/buckboost converter.