1

A little background first. I am trying to build a pretty dumb project where I try to take a VGA input and display it in a manner similar to a Nipkow disk.

The high frequency RGB signals need to be amplified greatly to output the 30V required for the 100W RGB LED, with each amplifier providing about 1/3 of this wattage for each color. This would need to produce relatively clean power at a very high frequency, around 100 or more MHz. I don't really care about efficiency that much at this point, I just want to get a working prototype.

I am not sure if building this kind of amplifier is within my low skill level I mean, I can't even solder properly yet, or if the costs will even be reasonable, considering I will probably need 3 (one for each color.)

I have little electronics knowledge, I only really know the bare basics. This is also my first stack exchange post, so give me criticism where it needs be.

Just a note: I think I have the electronics for the mechanical side sorted out, but I will post separate question if I need help there.

Before you ask: I want at least capable of 12 MHz as that is the required frequency to display a 300*200 at 200 Hz.

JRE
  • 67,678
  • 8
  • 104
  • 179
3eeeeeee512
  • 11
  • 3
  • Also, it's spelled solder, not soder. You're definitely going to need to solder for something like this--you can't do 100 MHz on a breadboard--so I suggest you learn that first. – Hearth Jun 16 '21 at 04:12
  • I know... I am just trying to get an idea of what I have to do. – 3eeeeeee512 Jun 16 '21 at 04:16
  • You don't need anywhere near 100 MHz -- maybe 100 KHz. – ErikR Jun 16 '21 at 04:17
  • 1
    Also, you'll need your amplifier's output to be a *current*, not a voltage--LED intensity is very nonlinear with voltage and very linear with current. Linearity is what you want. – Hearth Jun 16 '21 at 04:18
  • 100 KHz would only be sufficient to display about a 40*40 at 60Hz refresh... or in this case spin rate. I am hoping to get at least 300*200 at 200Hz... or 12 MHz... but higher is better I guess... it gives me more leeway in case I want to make improvements... like drilling even more than 200 holes in the rotating cylinder. – 3eeeeeee512 Jun 16 '21 at 04:23
  • This guy's youtube channel: [videolabguy](https://www.youtube.com/channel/UCnurlZAW7G0ybfGFNhJkYBg) has a bunch of videos on an analog implementation of a mechanically scanned "narrow band television" (NBTV) which can display a TV signal. For example [this video](https://www.youtube.com/watch?v=4UB9-FnBRAU) or search his channel for "NBTV". There's also [nbtv.org](http://www.nbtv.org/) which has a forum. – ErikR Jun 16 '21 at 05:32
  • LEDs have capacitance and power LEDs have a lot of capacitance. I don't know if your LED will be able to switch on/off at this frequency. – bobflux Jun 16 '21 at 06:55
  • @3eeeeeee512: Why 200Hz? That seems rather overkill given that movies and video are usually somewhere between 25 and 30Hz. – JRE Jun 16 '21 at 07:36
  • @ErikR Why is 100 kHz bandwidth enough? Standard VGA has a line rate of 31.47 kHz. 100 kHz would mean about 6 pixels per line. – Justme Jun 16 '21 at 08:17
  • @JRE the main reason why I want to make this project is to test high refresh rates without buying an expensive gaming monitor – 3eeeeeee512 Jun 16 '21 at 21:21
  • You should reconsider your project. Frequencies above 150kHz are considered radio, and without proper engineering skills and experience creating high power devices you can easily land into legal problems due to very possible interference to other radio devices if you will be successful creating an operational prototype in the first place. Not an easy task for an experienced professional. – Alexander R. Jul 06 '21 at 23:16

2 Answers2

1

You want a high-end video amplifier, it seems, similar to those that drove CRTs in the days of old. While some such amplifiers might still be available, the easy way to do this task is... a cascode amplifier. This is a pair of transistors, a 'lower' with a high frequency capability (but not high voltage), and a 'higher' with a high voltage capability (but not necesssarily high frequency). The general scheme also achieves high performance with a MOSFET or even vacuum tube (triode) for the 'higher'.

Example:

schematic

simulate this circuit – Schematic created using CircuitLab

There's a need to simulate all the parasitics of the components, including the light emitter, to get suitable gain, of course...

Driving an LED to high frequency is tricky: LED light comes from the charge carriers' recombination, which isn't always a fast process. The physics favors high frequency light modulation from lasers, instead.

Whit3rd
  • 7,177
  • 22
  • 27
-1

You can do that with an FPGA. You can get some FPGA dev board and play with it, and use some VGA IP libraries, if you are lucky you will find some.

A possible easier way would be to convert the VGA to HDMI, using some cheap converter and then interface that with an FPGA. There is already more resources on the web for that:

How to utilize HDMI port on FPGA (basic)

https://www.rs-online.com/designspark/hdmi-io-video-processing-system-on-the-artix-7-fpga-1

Once you get the HDMI decoded, you can simply control the LEDs using PWM output from the FPGA.

Controlling the LEDs in PWM is much simpler than by current, this will give the brightness impression by changing the DUT cycle of the PWM.

Some FPGA have a fairly high amount of IO, so you could control a few hundreds LEDs.

Although as a first electronic project, it might be quite difficult.

Damien
  • 7,827
  • 1
  • 12
  • 29
  • 1
    I don't think an FPGA is remotely necessary; this could be done entirely analog. – Hearth Jun 16 '21 at 04:40
  • I also think the interface does not matter. The problem of how to drive RGB Led brightness levels based on video data still exists. – Justme Jun 16 '21 at 04:43
  • @Hearth it certainly could, and deserves that you post an answer. – Damien Jun 16 '21 at 04:44
  • @Justme driving LEDs on a current basis is fairly easy. – Damien Jun 16 '21 at 04:45
  • @Damien That's a far more involved answer than I have the time or energy to do right now. That's practically a whole design. – Hearth Jun 16 '21 at 04:46
  • @Hearth That is why an FPGA based solution, when you basically uses libraries might be easier (but more expensive) than an analog design, with all the considerations required. There is also much more resources available online. – Damien Jun 16 '21 at 04:49
  • I am hoping to go solely analog, but inform me... I may be interested. – 3eeeeeee512 Jun 16 '21 at 04:51
  • Hah. You may think FPGAs are easy... I dislike even having to use microcontrollers if I can avoid it. Analog is so much easier to wrap one's head around. – Hearth Jun 16 '21 at 04:52
  • The thing is... varying the intensity of the LEDs using digital would either require converting back to analog or using some insane freqency PWM stuff (probably in the GHz) to actually get good color, which is why I am looking for just a design for a high power high frequency amplifier that is capable of driving LEDs. – 3eeeeeee512 Jun 16 '21 at 04:53
  • This is easily done in FPGA with DAC, so you can output analog on the pins, although the amount of LED will be limited to the available pins. Another way is to control the LEDs in digital using PWM. – Damien Jun 16 '21 at 04:57
  • @Damien I don't think you understand how a Nipkow disc works. There will only ever be three LEDs, but they need to be driven at incredibly high currents and the brightness needs to be adjusted incredibly fast. I would hate trying to do anything like that on an FPGA. – Hearth Jun 16 '21 at 04:59
  • The thing is... I am hoping to use vga both because I can test it on a crap computer that I don't care about first... and because I will be using dp to vga converters when I actually implement it. – 3eeeeeee512 Jun 16 '21 at 04:59
  • @Hearth PWM control for the brightness, a resistor to limit the current and a mosfet to switch it, there is nothing hard in controlling the LEDs. – Damien Jun 16 '21 at 05:01
  • @3eeeeeee512 simply use a cheap 10$ VGA to HDMI converter. – Damien Jun 16 '21 at 05:02
  • Yeah, you need a really fancy gate driver for that MOSFET. And the resistor better be a non-inductive type. And the FET needs to be quite large to handle that much current. – Hearth Jun 16 '21 at 05:02
  • @Hearth can always add a flywheel diode, and it's not related to the power but rather current, 100W LEDs are usually a string of LEDs in COB package, current is is like 3A at 30V, using a gate driver is also not something complicated. – Damien Jun 16 '21 at 05:05
  • @Damien the power is ~1a at ~30v – 3eeeeeee512 Jun 16 '21 at 05:06
  • @3eeeeeee512 then it's really no problem to drive with a N mosfet, even without gate driver. – Damien Jun 16 '21 at 05:07
  • @Damien Do remember that the question is asking about 100 MHz--your PWM frequency would have to be well over a GHz, and that means sub-nanosecond transition times on the FET's gate. There are very few if any gate drivers on the market that can manage that. Even if the FPGA can output it, which I doubt. – Hearth Jun 16 '21 at 05:08
  • @Hearth the drive frequency just needs to be above the eye response which is about 60Hz https://en.wikipedia.org/wiki/Flicker_fusion_threshold – Damien Jun 16 '21 at 05:09
  • @Damien Again, you're not understanding what a Nipkow disk is. – Hearth Jun 16 '21 at 05:10
  • @Hearth is it true that you could just vary the period of the signal during that 10 nanosecond window to vary the perceived intensity. – 3eeeeeee512 Jun 16 '21 at 05:10
  • @Hearth an FPGA can drive at few hundreds of MHz anyways and beyond that you have the LED response, especially if it's a phosphorus LEDs you'll get a few hundred us response. – Damien Jun 16 '21 at 05:11
  • @Damien the reason why I want to make this project is to test extremely high refresh rate monitors (above 200hz) without having to buy a fancy gaming monitor. – 3eeeeeee512 Jun 16 '21 at 05:12
  • How about an in-between. What if I could make a circuit that takes the 0-0.7v from the vga rgb and uses that to vary the period of a 100MHz pwm. Would that be possible? – 3eeeeeee512 Jun 16 '21 at 05:15
  • 1
    In any case, I'm not disputing that you *could* do it with an FPGA; I'm disputing the statement in the first sentence that you *need* an FPGA. You definitely don't need an FPGA, and trying to use one would be significantly harder than just trying to use a fast CFA as a transconductance amplifier. Or maybe you could even use an OTA, but I'm not sure if they come in high enough speeds. I've never had occasion to use one. – Hearth Jun 16 '21 at 05:15
  • Would a comparator amplifier work for this. I don't know much though. – 3eeeeeee512 Jun 16 '21 at 05:21
  • @3eeeeeee512 an analog design would require quite some skills at these frequencies, would require a custom PCB and I doubt you will be able to make it work if you have "little electronic knowledge", that is why I proposed FPGA as you can get dev boards and build using that. – Damien Jun 16 '21 at 05:23
  • @Damien so there isn't a relatively simple amplifier circuit that can amplify high power at high frequency? – 3eeeeeee512 Jun 16 '21 at 05:25
  • Can't FPGAs only output relatively low power, meaning you would still need an amplifier anyways? – 3eeeeeee512 Jun 16 '21 at 05:26
  • @3eeeeeee512 There are, but then you will have an analog output 0-30V, and you have the problem of driving the LEDs, you can't drive the LED with voltage so you need a whole analog circuit to drive it in current. It's totally doable, but will require quite some parts. – Damien Jun 16 '21 at 05:27
  • When you say FPGA I think of a dev board like arduino or raspberry pi... am I on the right track? – 3eeeeeee512 Jun 16 '21 at 05:27
  • @3eeeeeee512, no those are MCU or CPU, you need an FPGA (Xilinx) at these speeds. Driving with a low signal you use a PWM and a mosfet. – Damien Jun 16 '21 at 05:28
  • The way I understand a pwm circuit is that you take part of a sawtooth wave and use it to determine a period. Could a circuit be made that determines the "height" based on the 0-0.7v input, and therefore the period? – 3eeeeeee512 Jun 16 '21 at 05:29
  • @3eeeeeee512, Yes it can, but it would be better if you get familiar with electronics with a simpler project first. Breakdown this in parts and do them one by one. – Damien Jun 16 '21 at 05:34
  • I did a little more research and I found that the decay time of the phosphor would only allow for 2 MHz, which happens to be perfect for a 555 timer based pwm circuit right? – 3eeeeeee512 Jun 16 '21 at 05:37
  • @3eeeeeee512 breakdown your project and post specific questions – Damien Jun 16 '21 at 05:43
  • Ok, I will breakdown the project fully. – 3eeeeeee512 Jun 16 '21 at 05:45
  • What is being discussed here will be mainly important later in the project, which is finding a way to take a 0-0.7v input from a vga rgb source and using it to vary an led intensity at the frequency of the rgb source, with one led for each red, green, and blue. Like I said, I found that only about 2 MHz will be possible, due to phosphor decay time, but that is okay, this will be my first major electronics project. At first, I was thinking an analog amplififer, but now I think a PWM circuit will be best. That is pretty much all for the LED control stuff. – 3eeeeeee512 Jun 16 '21 at 05:50
  • The LED control will only be linked to the mechanical control through the computer itself. – 3eeeeeee512 Jun 16 '21 at 05:52
  • The mechanical side of this project is a bldc motor that spins at the same frequency as the vertical frequency asked by the computer. It will be controlled by an arduino, or in this case, an "elegoo uno r3" which is identical to the arduino uno r3, they use the same circuit. – 3eeeeeee512 Jun 16 '21 at 05:54
  • Here are the steps that the motor control takes – 3eeeeeee512 Jun 16 '21 at 05:54
  • first, the motor needs to spin at the same rps as the vertical refresh required. The esc takes a 50hz pwm signal with a 1ms minimum and 2ms maximum duty cycle to serve as the throttle. For each few duty cycles, the arduino will use a hall effect sensor, or optical sensor, not sure yet, both will do the same thing, the optical sensor would just require less balancing of the rotating cylinder, to determine the rotation speed of the motor. Relative to the required speed, it will increase or decrease throttle. – 3eeeeeee512 Jun 16 '21 at 05:59
  • I have this as a separate step to simplify things. The arduino will now use the hall effect sensor to increase the motor throttle for a split second, multiple times a second, until both vertical sync signals match. These vertical sync signals will be based on when a new frame starts, based on the many holes drilled in the cylinder in a raster pattern. – 3eeeeeee512 Jun 16 '21 at 06:02
  • The light will be shone through the cylinder, and, hopefully, an image will be produced. – 3eeeeeee512 Jun 16 '21 at 06:03
  • Don’t use a white LED, as the phosphor will take too long to dim; if you need white then combine red, green and blue LEDs, or just go with your favourite single colour initially. High power coloured COBs are rare but you can get reasonably powerful (3W) individual LEDs; use a small array of these as necessary. – Frog Jun 16 '21 at 07:26
  • @Frog I was going to use this https://www.amazon.com/gp/product/B01DBZIE4E?pf_rd_r=DHSVZPFHYWY14E8TWTYJ&pf_rd_p=5ae2c7f8-e0c6-4f35-9071-dc3240e894a8&pd_rd_r=958d3244-66d2-47af-a19a-31bacf500f2d&pd_rd_w=gZfHe&pd_rd_wg=R8EA4&ref_=pd_gw_unk&th=1 – 3eeeeeee512 Jun 16 '21 at 11:47
  • That looks quite impressive – Frog Jun 17 '21 at 05:36