How to build a 21st century protection board

[vzaichenko]

Hi All,

Here is my development on microcontroller-based "smart" soft-start / protection, or, I would say, amplifier control board (see attached).

Let's use this thread for discussions on firmware-driven boards - no matter what the PIC is, I believe, exchanging experiences will be useful for everyone interested.

Cheers,
Valery

P.S. My firmware and gerbers are available on request

[Krisfr]

Shucks you are turning me back into a programmer analyst, writing code and developing hardware AGAIN

But I am use to being that again, I want to focus on temperature and bias right at the output transistors, maybe mounting a LM35 right on the emitter or source pin right up next to the case, to read it temperature wise. While measuring the voltage drop across the emitter resistance, comparing them from device to device.

[vzaichenko]

Originally Posted by Krisfr

Shucks you are turning me back into a programmer analyst, writing code and developing hardware AGAIN

But I am use to being that again, I want to focus on temperature and bias right at the output transistors, maybe mounting a LM35 right on the emitter or source pin right up next to the case, to read it temperature wise. While measuring the voltage drop across the emitter resistance, comparing them from device to device.

Firmware-driven biasing is also an interesting one - didn't have time/courage to dig deeper, but it looks very attractive to me. When I thought about it some time ago, my initial idea was using some opto-pairs for bringing the current info to the controller (keeping it decoupled from the sensors at the same time)...

[ostripper]

I am deeply interested in this project - did you start with the "kit" on the
Arduino website ?

As far as the active biasing , I'm sure it would work ... "map an analog in ,
use a pwm output to brighten/dim a led". (this would be another
heavily filtered opto-coupler).

But , what would happen in the case of failure ? Thermal runaway !
The advantages could be sensing a thermaltrack diode or collector
pin SMD to update the micro in almost real time - still , analog backup
would be nice.

OS

[Krisfr]

What if we put together a demo board with one or two pair on it, a hundred watter or smaller to give everyone a chance to see how all this works. we could include a small prototyping area on the board to make things easy to add or changed. Maybe develop a schematic that could be easily bread boarded and shared, with lots of photos and code example. Kinda like the very good LTSpice thread that Molly has blessed the forum with. Lets start somewhere and let everyone find out that it is NOT hard to do at all, once you get past the basics. There are ton of UTUBE videos out there like I posted.

[ostripper]

Arduino = 16mhz clock - fast enough for basic bias adjustment. Also fast enough
for "on the fly" class A biasing with thermal feedback (safety).

I don't know if it could handle absolute real time class AB2 "fancy" biasing
or error correction. At this level , we would lose "analog" character ....
better off with class D.

It is not a PC , only 32K for any programming - (simple A/D -/set limits - then do X type operation).

OS

[vzaichenko]

Originally Posted by ostripper

I am deeply interested in this project - did you start with the "kit" on the
Arduino website ?

As far as the active biasing , I'm sure it would work ... "map an analog in ,
use a pwm output to brighten/dim a led". (this would be another
heavily filtered opto-coupler).

But , what would happen in the case of failure ? Thermal runaway !
The advantages could be sensing a thermaltrack diode or collector
pin SMD to update the micro in almost real time - still , analog backup
would be nice.

OS

Arduino environment, available at arduino.cc, does syntax "assistance" while writing the code, compilation and uploading. All you need.

Big advantage of such controllers - they are real time. Meaning - no operating system. The only thing running - is your program.

I don't like using PWM for such applications in audio - requires - as you say - heavy filtering and not reliable enough.
I prefer using a hardware DAC, either having it as part of the PIC chip (there are some), or having it as a separate chip. As an option - there is an appropriate Arduino extension board.

BTW, in some cases you may need more input/output pins, than Arduino can handle. In this case I use 74HC595N registers for extension and having the "locking" capability at the same time. They may be cascaded to have as much pins as you need (add up by 8 portions).
So, you push the desired "number" to the registers and lock it there. Then, before it's time to change the "number", you can do whatever you want in your program - handle protection, indication, etc. I used this approach in one of my professional designs, having one register to extend the inputs and two - to extend the outputs.

32K is A LOT. In the beginning of 80-th (last century ) I made my first computer myself (Intel 8080A processor, brought from Switzerland, 4.7MHz clock) very custom design, had to write CP-M core myself for it. That one had 16K RAM for everything... And no disks, so really for everything

Arduino is lightning fast, comparing to that one

P.S. I'm wrong - memory error - it was 2MHz clock, not 4.7

[vzaichenko]

Just had a power blackout in the whole building for 5 seconds.
AC power failure sensor triggered on the board, it disconnected the speakers immediately, then shut down the amplifier. PSU capacitors did not discharge in 5 seconds, so the board stayed alive, indicating "AC Failure" - one short flash every 2 seconds (Power LED).

Just as expected

[jwilhelm]

I'm playing with a slewmonster right now and am having some weird DC offset issues on startup. My protection seems to trip around 2 volts. Is that around the threshold it's supposed to happen at? This is kind of a good test for your circuit.

[vzaichenko]

Yes, this is around the threshold. Does it trigger?
Ah, ok - it does
Yes, I had it shutting down the amp a number of times as well during some dangerous experiments