Switching Flip-Dots current limited

[twam]

Hi everybody,

My next project involves switching a flip-dots matrix. This should be done 'pixel-wise'. Therefor I need a switch-able current-limited source which provides a define pulse which is then switched towards the correct flip-dot with some transistor arrays. The question is how to design this switch-able current-limited source with the following requirements:

• Should work with an input voltage of ~12 to ~24 V.
• Pulses need to be between 0.5 and 1 ms. This should be controllable by a µC, e.g. by having an enable input.
• Current should be limited to ~350 mA.
• High-Side

The first idea was to use e.g. a LM317 in current-limiting mode with an extra switching FET or a device like the FPF2702MX (http://www.onsemi.com/pub/Collateral/FPF2702-D.pdf) like done by https://github.com/545ch4/flippie. However my flip-dots have a coil resistance of ~18 ? which would mean that up to P=U*I=(24V-(18?*.35A))*.35A=17.7V*.35A=6.2W are dissipated in the current source. Therefor I think a switching current source would be much better.

The number of switching controllers/device is insane and it is hard for me to find a good candidate. Does any has a good suggestion where to start?

[cowana]

350mA sounds very much like the typical current LED drivers are designed for - have you explored switching LED drivers as a solution? For example, the PT4115 is a very cheap and common switching driver - it can cope with an input voltage of 6v to 30v, is high efficiency (up to 97%), and has the current set by a single resistor.

https://people.xiph.org/~xiphmont/thinkpad/PT4115E.pdf

[mikeselectricstuff]

As it's a short pulse, a simple resistive limiter should work fine as avarage dissipation won't be much.

[twam]

Thanks for the replies.

I had a quick look at some LED drivers, but wasn't aware of the PT4115 yet. However this seems to be a low-side switch. I'm not sure if this will work with my transistor arrays (TBD62783, http://www.mouser.com/ds/2/408/TBD62783APG_datasheet_en_20160511-775715.pdf and TBD62083, http://www.mouser.com/ds/2/408/TBD62083AFG_datasheet_en_20160511-769676.pdf) as they're directly connected to GND.

For calculating the resistive limiter I need to know the voltage supplied in advance, or did I miss something? Also the matrix consists of 14x16 dots and thus switching all of them with the minimum pulse length of 0.5 ms would take 112 ms. With a mechanical switching time of max. 70 ms per dot I would basically always switch a dot to get maximum FPS on the display and therefor I think that the average dissipation is far more negligible.

[cowana]

The low side transistors makes things slightly more complex.

Even if you did use a high side switch such as the TI LM3405 (http://www.ti.com/lit/ds/symlink/lm3405.pdf), they typically reference their sense resistor to ground - as the feedback voltage is 0.205v, if you had the low side switching matrix below that the transistor's drop would be interpreted as a very large current flowing.

It may be possible to float the low side switching matrix to the feedback voltage (0.2v)?

[mikeselectricstuff]

I'm sure there must be some LED drivers with high-side current-sense.


Also bear in mind that flipdots are an inductive load - this may upset LED drivers, but may also help in that the current won;t go to 100% immediately so the avarage current over a pulse may be significantly less than its resistance would imply - do some measurements.

Another option would be to use a buck voltage reg down to a constant voltage and use a resistive limiter from that.

[Zero999]

Hi everybody,

My next project involves switching a flip-dots matrix. This should be done 'pixel-wise'. Therefor I need a switch-able current-limited source which provides a define pulse which is then switched towards the correct flip-dot with some transistor arrays. The question is how to design this switch-able current-limited source with the following requirements:

• Should work with an input voltage of ~12 to ~24 V.
• Pulses need to be between 0.5 and 1 ms. This should be controllable by a µC, e.g. by having an enable input.
• Current should be limited to ~350 mA.
• High-Side

The first idea was to use e.g. a LM317 in current-limiting mode with an extra switching FET or a device like the FPF2702MX (http://www.onsemi.com/pub/Collateral/FPF2702-D.pdf) like done by https://github.com/545ch4/flippie. However my flip-dots have a coil resistance of ~18 ? which would mean that up to P=U*I=(24V-(18?*.35A))*.35A=17.7V*.35A=6.2W are dissipated in the current source. Therefor I think a switching current source would be much better.

The number of switching controllers/device is insane and it is hard for me to find a good candidate. Does any has a good suggestion where to start?

How often is the dot flipped? I presume they're latching and remain in the same state, without consuming additonal power. If the duty cycle is low, then the average power dissipated by a linear regulator, could be quite low. The problem with an LM317 is it requires about 4V of headroom, when used as a current regulator.

The coils are inductive, so it will take time for the current to build up. All that's needed is a monostable circuit which uses the inductance of the coil as the timing element: wait for a trigger pulse, apply power to the coil, monitor the current, until it exceeds a certain threshold (high enough to flip the dot), then cut the power to the coil.

[floobydust]

Original flip-dots circuits use capacitive-discharge instead of constant-current.
It's so your power supply load is less and no smoke if a driver gets stuck on due to bad firmware or shorted transistor.
If you run say 20VDC you can also use narrower pulses to flip the dots.

Slowly charge a big cap up to high voltage and use that to hit the coils.
Like capacitive-discharge model railroad switchups.