Custom sig gen to integrate in a project

[stcoso]

Hi!


I've recently designed and built a LED driver (based on a Howland current source). It works well but it (obviously) needs a signal generator to be driven.

This are the requirements:

• Only positive square waves (and positive DC)
• Reasonable rise/fall times (25/30ns)
• Frequency up to 1MHz
• Variable duty cycle. Obviously at 1MHz i do not expect to be able to go to 1%   
• 0-10Vpp but i'll need an amplifing stage so this is not a big issue

So... a DAC? a DDS ic? An MCU with integated DAC? Something else???

Thank you in advance

[mvs]

Pulse wave is not essentially analog signal, so why to use DAC or even DDS?
MCUs with timer/counter are usually able to output pulse wave. Frequency and resolution will depend on timer clock speed. Some AVR MCUs have 64 MHz timers, more modern ARM MCUs can go over 100MHz.
To amplify signal to 10Vpp you can use some fast comparator.

[stcoso]

Because I need to some control over the amplitude

[mvs]

You will need quite fast DAC to achive 25/30ns reise/fall times. They are usually not cheap.

You can try to build switchable attenuator or R2R ladder DAC using MCU GPIOs and resistors. To amplify signal you will need a fast opamp with slew rate >300V/µs.
But all this effort just to make a LED driver work, is a bit ridiculous...

[stcoso]

Before judging one should know WHY the effort... just saying.

[Terry Bites]

Look for high speed PGA and VGA devcies eg VCA822 to drive your current source.
Feed your pulses through a high speed comparator to generate fast rise times. eg LTC6752- you may want an attenuator inbetween the comparator and the VGA
Make sure you read app notes on high speed comparators- they can be fussy.
Good pcb layout and decoupling is a must if you are dealing with fast edges.
You'll be going no where on a breadboard. Protype it manhattan style.
Loads of choices for comparator and PGA/ VGA. form AD and TI and others.

[srb1954]

I've recently designed and built a LED driver (based on a Howland current source). It works well but it (obviously) needs a signal generator to be driven.

This are the requirements:

• Only positive square waves (and positive DC)
• Reasonable rise/fall times (25/30ns)
• Frequency up to 1MHz
• Variable duty cycle. Obviously at 1MHz i do not expect to be able to go to 1%   
• 0-10Vpp but i'll need an amplifing stage so this is not a big issue

Why do you need 1Mhz frequency and 25ns rise/fall times?

Does your Howland current source circuit even respond to signals this fast?

[Terry Bites]

I've worked with fast current sources for years I'd be truly amazed if any one can get submicrosecond resposes from one. The parasitics will through your caerfully selected resistors in to imbalance.
What output impedance are you aiming at?
I'd love to see your shematic for it.
If you need a fast high current pulse in an  LED, dump a capacitors charge through an inductor.
https://www.sciencedirect.com/science/article/abs/pii/S1005888516600237

[ogden]

Because I need to some control over the amplitude

Control supply voltage of level converter. Then you don't need DAC with fast settling time, any cheap I2C DAC will do the job.

[Terry Bites]

Lets see the schematic so we can judge if your're input speed requiremets are actually necessary.

[stcoso]

I've recently designed and built a LED driver (based on a Howland current source). It works well but it (obviously) needs a signal generator to be driven.

This are the requirements:

• Only positive square waves (and positive DC)
• Reasonable rise/fall times (25/30ns)
• Frequency up to 1MHz
• Variable duty cycle. Obviously at 1MHz i do not expect to be able to go to 1%   
• 0-10Vpp but i'll need an amplifing stage so this is not a big issue

Why do you need 1Mhz frequency and 25ns rise/fall times?

Does your Howland current source circuit even respond to signals this fast?

I'm sorry for the delay in my responses, I've had too many things to manage. Hope this will not be considered necroposting.


Anyway, yes, it does respond quiet well and it's nothing too exotic. It's really similar to the "CAEHCS" illustrated here: https://www.analog.com/en/analog-dialogue/articles/a-large-current-source-with-high-accuracy-and-fast-settling.html

[stcoso]

I've worked with fast current sources for years I'd be truly amazed if any one can get submicrosecond resposes from one. The parasitics will through your caerfully selected resistors in to imbalance.
What output impedance are you aiming at?
I'd love to see your shematic for it.
If you need a fast high current pulse in an  LED, dump a capacitors charge through an inductor.
https://www.sciencedirect.com/science/article/abs/pii/S1005888516600237

Cannot use this (classic) approach. I need (optical output) square wave.

[stcoso]

Lets see the schematic so we can judge if your're input speed requiremets are actually necessary.

I do not really understand this attitude... in anycase the schematic is really similar to the one posted above from analog.com . I'm not able to show right now proofs that it works but if really needed I've a video of an early prototype from 2020 driving a LED@1A (500khz duty ~25%) with already decent rise/fall times. The measure was done with a photodiode terminated directly into the scope 50Ohm. I'm attaching a screenshot...