Why no error amplifiers in constant off time LED drivers?

[Flyback]

The LM3421 is the only constant off time LED driver IC in the world which has an internal error amplifier.
Why is there only one such IC in the world?

LM3421 DATASHEET:
https://www.electro-tech-online.com/custompdfs/2013/09/lm3421.pdf

Constant off time is THE answer to driving long series strings of LEDs from low input voltages, because 'constant off time' lets you do high step up ratios without needing slope compensation, and is devoid of the maximum duty cycle limitations which are prevalent in most constant frequency PWM controllers.

The HV9910B is a constant off time controller , but does not contain in internal error amplifier.

HV9910B datasheet:
https://www.electro-tech-online.com/custompdfs/2013/09/HV9910B.pdf

 

[MrAl]

Hi,

There are a number of chips that use this technique or a similar, simpler technique. The main idea is to pump the inductor with enough energy so that when it discharges it provides enough of a boost to get to the required level of voltage at the required current and that is able to happen when the inductor gets to a preset current level. There is feedback, but it's not the typical error amp loop like we normally see because it is assumed that the regulation control does not have to be that accurate and that several operating parameters are nearly constant.

The problem of course is that there is no regulating feedback, only inductor current limiting. Without feedback the circuit will not regulate very well when external parameters change such as load and input voltage.

There are other types of control however, that do not involve feedback but are actually feed forward. That is, they measure something about the circuit before the control stage and vary the control stage based on assumptions about the circuit such as power losses. In the case of the max current controlled inductor scheme, measuring the input voltage allows the circuit to change the control to the inductor to at least partially make up for the changing input. This can make a big difference because now we can adjust the output based on input voltage, at least to some degree of accuracy, without having to directly measure the output itself. This eliminates the feedback loop and thus the problem with the right hand plane pole and other problems that plague boost converters. The drawback is the control is not as accurate so it is best for loads that do not require super accurate current or voltage levels to operate properly.

The first chip you linked to appears to be using this kind of scheme. I have not looked at the second link yet.

So you see this kind of controller is best for loads that do not require super accurate outputs like LED drivers.

 

[Flyback]

Thanks, you describe the HV9910B well and informatively, the LM3421 , however, as you know, does have an internal error amplifier and can very accurately regulate led current in spite of line and load changes.
The problem is, the LM3421 is the only constant off time chip in the world that incorporates an internal error amplifier.

 

[MrAl]

Hello again,

Oh so you meant that it uses constant off time but it also has an internal error amp. I did not read the data sheet in it's entirety but i did read that is uses feed forward also.

So you need another chip that does the same thing, or has an error amp built it?

 

[Flyback]

yes, I need a chip that's simple like HV9910B but with a built in error amplifier. The LM3421 is good but restricted, for example, you couldn't use LM3421 in buck boost with vin = >75V.

 

[alec_t]

Would it be a big deal to use an external error amplifier? That would give you a wider choice of chip.

 

[Flyback]

Would it be a big deal to use an externall error amplifier? That would give you a wider choice of chip

I know what you mean, in fact, on the last design, I did use an external error amplifier. However, I find bosses never like extra components, and so it'd be nice to find something with the error amplifier inside.
The next design I have in mind is for a xenon flashlamp charger circuit (charging the HV cap up repeatedly.)......and its done using few components as it is, and I wish to convert it to using an ic, as the discrete version that it is now has certain problems....however, its very cheap, so the boss wont like the replacement circuit having an extra external amp added as an error amp.

 

[ronsimpson]

1) Why do you need very accurate LED current? I don't think your eye can see 20% changes in current.

2)It appears that you will be using a long string of LEDs so the LED voltage is high. You will not loose much power with a current sense resistor. Many PWM are built to regulate at 2.5V. There are some at 1.25 volts. Out of 75 volts that is not much loss. Some regulators, regulate at 0.8V or lower voltage.

3) The HV9910 will hold the LED current pretty close with changes in input voltage. The MOSFET will open up when the inductor current causes 0.25 volts across the current sense resistor. My guess is that you will have 1 or 2% regulation.