balanced headphone protector

[Pars]

Have a version of this breadboarded right now and something doesn't seem right. I was going to modify this to put in a manual mute switch and a pair of LEDs.

 

I only have a pair of inputs wired up to the LM339 (single right now), and at the junction of the 1K/100K/500K resistors and the cap, I am seeing -9V or so, feeding the gate of the FET. Relay is off. Since the relay is connected to signal using the NO contacts, I should think with no signal in, it should be ON and conducting the input signals to the output?

 

I am using a 47uf electrolytic cap currently, with the - on V- and the + at the resistor juncture. I do not have the protection diodes in on the 2 inputs I have wired. The outputs of the LM339 are sitting at ~ V- unless I overload using a 9V battery. I've never used the LM339 before, particularly on dual supplies.

[GrindingThud]

Single ended input (negative grounded) I get +7.9V at that junction. There are two LM339 in the circuit and both are wired. Takes a few seconds to ramp up from around -10V and trigger the relay on.

Pins 1, 2, 13, and 14 of the LM339 should all be at 9V or so when output relay on.

Edited April 11, 2015 by GrindingThud

[kevin gilmore]

if you have open inputs on the lm339 the bias current will turn them on.

 

need to make sure the inputs before the filters are tied to ground.

 

+7.9 is the correct voltage wrt ground when the relay is on.

[Pars]

I only had 2 of the outputs tied together, but the inputs were floating (just a single 100k resistor). I'll tie both of these to ground.

 

I am running this from a +/- bench supply and noticed the 100k/500k combo only ohmed out at 8k or so in series with the bench supply connected. Removing it and the resistance went up to 22k or so, which makes sense since the reference voltage divider is in parallel and is 23k.

 

I may dig up some 12V regulators and wire them up to isolate the circuit from the bench supply. I also threw a 10uf film cap in in place of the electrolytic to no avail.

[Pars]

Got this working, think it was floating input causing the problem.

 

I dug up a pair of 12V regulators and put them in, but no joy. Turns out the positive reg was an ST Micro (probably one of the pair that I had problems with) and only put out 11.05V (vs. -11.9+V on the neg regulator, not ST Micro), which had the effect of unbalancing the reference voltage divider. Ref voltages were +0.05V and -1.01V, which didn't work so well

 

Went back to just the bench PSU and it started working. I'll have to grab some more 12V regulators... have all sorts of 15V etc. but no 12V other than LM317/337s.

[kevin gilmore]

you should be able to use 15v regulators, the relay will get slightly hotter

[Pars]

I just have it breadboarded to test it and the mute switch/LED additions. I'll do an actual PCB up (mostly done) when I verify a couple of things there. And will use 12V regs on the board.

 

I do have some of the TO92 regs (78L12/79L12) but assumed these might not be the best thermally with the relay being driven off the + one.

[gepardcv]

I just tried adding this circuit to my in-progress supersymmetry Dynalo build. First few times I powered it up, I thought I heard the relay clicking, and there was sound on the output. Then the clicking stopped, and I don't get output anymore. The first few times I powered it up, I was fixing wiring mistakes with the amp boards, but I'm pretty sure they're fixed now. Question: is there a simple way to test if I fried the board? I don't see char marks anywhere. The amp boards seem to be fine, since I get sound when I wire their outputs directly to the headphone outputs.

[Pars]

Check the voltage + and - references at the 1K resistor in the voltage divider comprised of 11K / 1K / 11K. You should have +0.5V for the + reference and -0.5V for the - reference. You can check these on pin 7 on U1 (+) and pin 8 of U2 (-). Also check the +/- 12V.

 

If the comparator trips for a fault, you will see around -12V (-11.x) at the junction of the 100K / 500K resistor (relay off). If all is fine, this will be ~ +7.x Vdc (relay on).

 

All measurements to ground.

[gepardcv]

0.005 V (basically 0) on both pin 7 of U1 and pin 8 of U2. -12 V at the 500k resistor. I should just make another of these boards because this one is fried, yes?

[Pars]

Do you have +12V (top of the 100K resistor, pin3 on U1/U2)?

[gepardcv]

Yes, and here's something odd: I just powered up the amp with all inputs and outputs disconnected from the protect3 board, and the relay clicked. Maybe the board is fine, but I still have a wiring mistake somewhere and it causes a fault.

[gepardcv]

Clarification: I just observed the following:

 

1. With nothing connected, the relay clicks about 4-5 seconds after power on. Pins 3 on U1 and U2 show +13.2V before the click, and +12.8V after the click. The 500k resistor always shows -12V, and never flips.

 

2. With anything connected on the input side, the relay never clicks.

[Pars]

Make sure the 11K resistors and the 1K are correct, and measure the voltage drops across them. Something isn't right with your reference voltages of 5mV was it? Those should both be around 0.5V, one + and the other -.

[gepardcv]

Something is weird with the resistors. The 11k and 1k look fine, but the 100k and the 500k show something crazy, multiple M, then drop and fluctuate. I think I'll just order some more parts and rebuild it. Plenty of boards, luckily.

[Pars]

Remember that the 47uf cap will effect what you are seeing when you ohm out the 100k and 500k resistors. Add the fact that these are in parallel with the 11k/1k/11k string and you will see some things that initially will look strange when ohming them out. I doubt there is anything unfixable with your current board, but do as you wish.

 

The problem you are seeing relates to your very low reference voltages of 5mV, which is what the comparators are basing their decisions on DC offset, so if these aren't right, the circuit will not work. Also, the + and - 12V from the regulators needs to be pretty balanced. I had a bad + regulator (11.05V, with around -11.95V on the other rail) and was seeing +0.005V and -1.05V references, which of course didn't work.

[gepardcv]

Turns out I soldered the 7812 backwards. /facepalm

 

It works now.

 

Thanks for helping debug this thing!

[Pars]

[Pars]

I finished up my breadboard testing for adding a manual mute switch and status LEDs to this circuit. I wound up going back to the 2N7000 with a PNP as used on amb's epsilon12 circuit. I could not get the LEDs to work properly on everything else that I tried, as below:

 

1. Original circuit, with mute switch between relay coil and 2N7000 drain. Muted fine, but not enough voltage to drive the LED circuit.
2. Moved the 2N7000 to above the relay, in between +12V and relay coil. This would not energize the relay.

 

Here is the schematic of what I am planning on using.

 

 

And here is the resulting board layout, extensively cribbed from Dr. Gilmore's layout (though I did try some others first)

 

 

Still working on label clean-up, etc. Also, the BC 5xx transistors I used could probably be replaced with 2N3904 / 2N3906; just flip them around. I noted the transistor pin-outs on the layout, though they become slightly misaligned using Seeed Studio's CAM processor job.

 

I also put in both the Phoenix terminal blocks and Molex KK connectors. The mute switch and LEDs only have the Molex connectors currently, with a shared ground on the LED connector. I think I will switch to a pair of 2-pin or a single 4-pin connector and might add the Phoenix terminal blocks for those if anyone is interested. Board size would probably go up a bit with those as well.

Edited April 20, 2015 by Pars

[gepardcv]

This is really cool!

[Pars]

I ordered some of the above boards, have built one up with parts on hand, and everything seems to be working. I am trying to test the protection function currently, and have all inputs but 1 grounded. I am injecting offset into L+ input with a 9V battery, and can't get it to trip. Verified at the LM339 pins as +8.x Vdc, with +0.47 and -0.53V reference voltages.

pin 6: -ref

pin 7: 8V

pin 8: 8V

pin 9:+ref

Am I misunderstanding this or not a good test scenario for this?

[kevin gilmore]

I would make sure the comparitors are working as expected. .1hz sine wave makes it easy

[Pars]

Putting the 2N7000 in the right way seemed to help

These comparators do work differently than one might expect, but it seems to be working fine now.

[kevin gilmore]

bugs bunny:  Oops i had the silly thing in reverse

[wink]

Sounds like my car......