How to build a 21st century protection board

[vzaichenko]

A question regarding the thermal detection with the MJE340's - how have people mounted them with respect to placement on the heatsink?

In my case with 3 output pairs per channel, is there a common school of thought as to where is an effective monitoring position?
I've got a piece of aluminium flat spreader common to my transistors, is this a reasonable mounting site?

The question in the back of my mind is what if one device is running hot, but it's not enough to raise the total temp to the overheat threshold....

I guess the answer would be that this is a device failure issue which the current sensing would catch? Which, if I'm on the right track would suggest that the over current protection trigger should be for the whole string of output devices rather than a single pair?

Does this sound right or have I wandered off target? 🙂

Hi UKsuperdude,

You're thinking in the right direction 😎

Basic principles behind the board functionality are:
- Overcurrent protection watches short time current spikes above the threshold;
- Overtemperature channel watches much slower, heavily integrated temperature rise.

Normally, situation of overtemperature may happen in case the amp is running at high power for a long time. It still runs fine, but the temperature of the output devices becomes dangerously high. In "professional" configurations, like stage amplifiers, we can use more complicated setup, controlling the cooling fans in the chassis. In a simple "home system" setup we just shut down the amp.

So, the best position of MJE340 temperature sensors on the heatsinks - somewhere close to the area where the output devices are mounted. We use 2 sensors, assuming we monitor the left and right heatsinks' temperature in a frequently used compartment configuration.

Overcurrent - we monitor one pair of output devices, assuming the current distribution is even enough between the pairs in a well-designed and carefully-built amplifier. But still, even if, say, one of the transistors fails momentarily, shorting the output to one of the rails, it causes causes high current spike in the opposite shoulder of the output stage. And even if this spike is not enough for overcurrent sensor to trip the protection, DC offset sensor will do it for sure.

So, having those 3 protection channels - overtemperature, overcurrent and DC offset - working as an integrated system, ensures high reliability of the board, protecting both the amplifier and the speakers.

Cheers,
Valery

 

[evanc]

I put together another protection set last night. Found out that the atmega chips I purchased did not have the bootloader installed. Luckily I had one older chip that was ready to go. (I'll work on this problem later)
When I started the set without hooking it to the rails as expected it flashed DC offset. Once the rails were connected the set flashes once every two seconds....AC fault. When I check the board has 11.94 volts. It was late last night so I put off trouble shooting until today. I'll start by checking part values. Any hints as to where it would be good to look. The other sets I have put together worked right off.
Thanks,
Evan

 

[evanc]

Here are the schematics I followed.

 

[jwilhelm]

Check your voltage across D4. That should be around 5.1 V. If it's around there, U7 should be pulling it's pin 3 low, which in turn should pull pin 17 on the Atmega328 low.

Installing the bootloader is fairly painless. You can use your Uno as a programmer, and a control board as a fixture for your Atmega328 for programming. You only need to install the Atmega328, R1, and the crystal for programming. You can't use a complete control board for this because there are pull up resistors on the data lines you need to use. https://www.arduino.cc/en/Tutorial/ArduinoISP

 

[evanc]

only 4.5 volts across D4. This morning the amp starts up...no fault light.

 

[evanc]

I take it back...Less then 3 volts across D4 ac fault blinking. Checking at J1 There is 5 volts referenced to ground.

 

[jwilhelm]

That circuit makes it's own DC from one of the transformer AC connections. If all connections look good, try increasing the value of C2.

 

[evanc]

looks simple enough. It seems weird that the voltage is sometimes correct and sometimes not. I'll check/change the few parts involved and report back.
Thanks,
E

 

[jwilhelm]

The weird voltage actually sounds like a cold solder joint of flaky transformer connection. The circuit is doing it's intended job. It has it's own supply so the reservoir caps won't slow it's reaction time down.

 

[evanc]

First I reflowed the solder joints one at a time testing after each. No difference. Replaced D4 and now it seems to work. I cycled the amp about 20 times and it works as it's supposed to. Thanks for the hand holding.
E

 

[jwilhelm]

So a bad zener. Is D4 getting warm?

 

[evanc]

doesn't feel warm. Maybe I held the iron on the first one too long..Not sure.

 

[jwilhelm]

It's working. That's the important thing. Next you get to learn how to install a bootloader.😀

 

[tauro0221]

Hi,
I found this AD5592R chip very interesting for those that are using the Arduino and ran out of I/O ports. Attached it is the datasheet for your consideration/evaluation. I posted here since this thread it is about the used of micro for controlling.

 

[jwilhelm]

Hi Tauro.

The AD5593R looks like a good option for I2C interface too. That looks like it might be handy.

 

[evanc]

OK, I did a little more measuring and adjusting and although the protection circuits are working the voltages and current readings don't seem to be in line with what I expected. I apologize for not being able to work these things out on my own. All 4 boards are acting the same.

On the dc detect/speaker protect board
Current across R7 was low. I see 58 volts here and adjusted the resistor value to 2K to get29 mA

R8 was 10.4V across 750R for 13.9 mA. I adjusted to 428R and now voltage has dropped to 9.4V for 22mA

R9 was 2.4V across 430R for 5.58mA. I adjusted to 186R and voltage dropped to 1.6V for 10mA. I am afraid to continue dropping the resistance without guidance.

The regulators on the control board are putting out correct voltages.
Thanks,
E

 

[jwilhelm]

U1 can actually run at lower current. It's fine at 16mA, so the current through R7 can be decreased as well to around 20mA. I've found R7 will get pretty hot with higher rail voltages at 30mA.

What is your collector to emitter voltage on Q9 and voltage across R11?

 

[evanc]

Q9 VCE is 4.17V across R11 .61V

 

[jwilhelm]

Q9 isn't turning on fully. Drop R11 value to 6k8 of 5k6.

 

[evanc]

Well while probing around the live circuit....you see it coming....I momentarily shorted one of the rails to ground at the speaker protect board. Bright flash loud bang and a blackened probe. That seems to be the extent of it. Recycled the amp and it seems to work OK.
readings now:
R7 2.2K 57V 26mA
R8 428R 9.4V 21mA
R9 187R 1.65V 8.82mA
R11 6.8K .37V .05mA Same current as 12K
Q9 VCE 3.8V

If need be I can build another, not blackened, protection board....but not tonight....