Amplifier Rail Detector - Protection Circuit

[thylantyr]

Need some ideas on how to make this circuit better.
I slapped it together on circuit maker, it's an untested
circuit.

Circuit goals;

1. To detect a large differential in voltage between
both positive and negative rails on your amplifier's
power supply. The attached schematic is set for
8v differential.

2. To detect if one of two rail fuses blows.

3. Operation of the circuit should be rail voltage
independent as much as possible. It should operate
with low voltage rails or high voltage rails. Example
is modeled with +100v/-100v rails.

4. The circuit will output a ground when a fault is detected,
otherwise the circuit pulls up to +12vdc.


There is two complementary circuits, one to detect pos rail,
the other to detect neg rails. The circuit triggers when
there is a 8v differential between rails. This threshold is
adjustable by replacing the zeners D1 and D2 with desired
ones.

The output of the positive detector is -12v (untriggered)
and ground when triggered.


The output of the negative detector is +12v (untriggered)
and ground when triggered.


How do I "Or gate" these two opposite output voltages
so it operates on a single +12v supply ?



My silly solution was the Q3, D3, D4 circuit with
the Zener D5, D6 (+12v, -12v references). D3 is
used to clamp the negative trigger voltage closer
to ground.

Is there a bettery way?

*******************

Input circuit description;
R1, D1, R3, Q1 detect if there is a neg voltage rail failure.

R2, D2, R4, Q2 detect if there is a pos voltage fail falure.

C1 is used to smooth the bumpy DC rails.
C2, C3 are there to minimize noise on the base.

If both rails are within 8v of each other, the circuit
doesn't trigger. D1, D2 creates the threhold voltage of
triggering. Increase Zener voltage to increase window
and visa versa.

This is one of many protection circuit modules I want
to incorporate into my plans. When this circuit outputs
the ground trigger, the relay module will disable the
relay connected to the speakers.

It's NOT tested yet! It's my first circuit design attempt
in 10 years

[thylantyr]

All is quiet on the forum

I simplified and deleted 1 resistor and 1 zener from
the design. I think I'm done thinking about this,
time to build one to test it.

Oh.. forgot to mention, C1 also determines
the reaction time, big C1 = more delay.
I think it needs to be smaller, perhaps 1uf.

Changing topics, any interesting methods
to detect output short circuit on the amplifier
output (output stage), a short to either rails
or ground

The common solution is the series resistor and
transistor for short circuit detection, but is
there another method withing using the series
resistors?

[joensd]

Hi, have a look at this protection circuit : Problems with Crescendo ME - Need help!
The first page holds the schematic for the amp, the second the protection circuitry.
It actually has all the features you describe including output overload (Currents in excess of 5A).

Regards
Jens

[thylantyr]

Originally posted by joensd
Hi, have a look at this protection circuit : Problems with Crescendo ME - Need help!
The first page holds the schematic for the amp, the second the protection circuitry.
It actually has all the features you describe including output overload (Currents in excess of 5A).

Regards
Jens

Funner to build your own

.................................................. .

Any recommendations on
which high voltage transistors
to use ?

Does anyone see any issues I
overlooked ?

[theChris]

btw, thy, do you think this would work?

(see included circut)

i would guess you could use bipolars driving the LEDs instead of just LEDs. it should compare ground to V1/k - V2/k.

this also reduces the need for high voltage ratings, right?

[janneman]

Originally posted by thylantyr


Funner to build your own

.................................................. .

[snip]

Yes, it is more fun to design your own

This is how I would do it:
Make a +/- supply for a couple of opamps (+/- 5v should do with a couple of zeners).
Scale both the supplies to be protected to say a couple of volts in normal operation. The pos voltage goes to an opamp input.
Another opamp getting the neg pol input, is a unity gain inverter. Its output goes to the other input of the first opamp. This first opamp must have a gain depending on the scaling mentioned before, of course. This output is a measure of the difference between the two supplies.
There is a third opamp running open loop with just a touch of pos feedback to give it some hysteresis for clean switching. This last opamp has a ref voltage on one input, and the output of the first opamp at the other. It will swich when the first opamp output is over/under the ref voltage.

My 2 eurocents (2.17 dollarcents) worth.

Jan Didden.

[thylantyr]

Originally posted by theChris
btw, thy, do you think this would work?

(see included circut)

i would guess you could use bipolars driving the LEDs instead of just LEDs. it should compare ground to V1/k - V2/k.

this also reduces the need for high voltage ratings, right?

Referencing your pic, short both R2's, delete both R3's
and you have my input stage which swings +5v or -5v
on a 10v differential. Your extra stages swings +2.5v and -2.5v
which works too. All you need to do is convert those voltage
swings to +12v out (no fault detected) to ground output
when a fault is detected. This circuit needs to drive
the controller board which needs a ground input to trigger
and +12v pull ups.

protection circuit #1 ---> controller input #1
protection circuit #2 ---> controller input #2
protection circuit #3 ---> controller input #3
protection circuit #4 ---> controller input #4
etc.

I'm trying to figure out each protection circuit
independently, then feed each protection module
into the main controller board which drives the
speaker relay. It's easier for me to debug and test
independent circuits plus have the option to enable
or disable certain protections depending on
what you are doing with the amplifier.

I'm not doing anything new, I just want to break it
down into modules.

[thylantyr]

Originally posted by janneman


Yes, it is more fun to design your own

This is how I would do it:
Make a +/- supply for a couple of opamps (+/- 5v should do with a couple of zeners).
Scale both the supplies to be protected to say a couple of volts in normal operation. The pos voltage goes to an opamp input.
Another opamp getting the neg pol input, is a unity gain inverter. Its output goes to the other input of the first opamp. This first opamp must have a gain depending on the scaling mentioned before, of course. This output is a measure of the difference between the two supplies.
There is a third opamp running open loop with just a touch of pos feedback to give it some hysteresis for clean switching. This last opamp has a ref voltage on one input, and the output of the first opamp at the other. It will swich when the first opamp output is over/under the ref voltage.

My 2 eurocents (2.17 dollarcents) worth.

Jan Didden.

Thanks for the tip. I will check it out. One of my goals was to explore a non-opamp solution then, later try to figure
out the same circuit using op-amps to see which one
yields less PCB space.

[thylantyr]

60 downloads...

Anyways, I'm going to build the original circuit within
a few weeks, ordering parts. I'm gonna use MPSA92/42
transistors, HV overkill rated. heh

I think the original circuit is flawed. Too much reverse
voltage and the base according the circuitmaker,
so I changed the input resistors to 10k going into
the two transistors and added a generic diode
at the base (to ground) to clamp the voltage.

I'm also gonna build JD's op-amp version, I need to decypher
the cryptic post, I'm no analog circuit designer so beware
of my crazy ideas.

[thylantyr]

Uploading the latest revision to
the circuit. It was partially tested.

Some changes were made from
theory to real testing.

I did some partial tests with variable
rails from +12/-12 to +20/-20v
and the circuit responded.

But, I didn't use the 3.3v zeners,
instead I inserted a short where
the zeners would be. The circuit
triggered when rail voltage
differential exceeded 2 volts
(somewhere between 2v-3v actually).

Zener testing and real amplifier tests
will be soon.

Circuitmaker is a pretty cool
tool.