[amb]

I present to you the ε12 ("epsilon 12"), a turn-on delay muting and DC offset protection circuit designed specifically for headphone amps. It prevents the turn-on "thump" and turn-off noise present with some amps, as well as provide protection to the headphones in case the amp's output DC offset goes awry during a malfunction.

The ε12 can be used with a traditional 2-channel amp, a three-channel amp (M³, PPA, Pimeta, etc.), or a four-channel amp (fully balanced stereo outputs), and will provide for DC offset protection on all available channels. This circuit is intended to be powered from the amp's DC rails (up to +/-30V) and has its own voltage regulation and virtual ground reference, which is adjustable.

The headphone outputs from the amp is routed through a relay, which keeps the headphones disconnected until after ~3 seconds delay after power up. The circuit will disconnect the headphones if any of the channels being monitored produces in excess of 70mV or DC offset (positive or negative), and will reconnect the headphones if the offset drops to a safe threshold. The 70mV trigger point is chosen to be sensitive enough to protect most headphones from damage under normal circumstances, but not too sensitive to get false triggering. Compare this to many DC offset protection circuits for speaker amps, which would not trigger until over 600mV. The relay as specified is a high quality Tyco/P&B RT series unit with 90-10 silver-nickle contacts rated at 8A, but you can use another 12V DC relay as long as the coil current comparable.

The adjustable virtual ground reference pot let's you set the "zero" reference upon which all channels are compared.

The circuit is designed to be simple, low in parts count and cost, yet provide all the functionality desired.

I have a working prototype on breadboard, and I invite your feedback. If you wish to build the circuit, feel free to make your own PCB or use Point-to-point wiring on a perfboard. I have no immediate plans to offer pre-made PCBs, unless there is sufficient demand to justify it.

The name "ε12" has no special significance, other than that I tend to name my DIY projects with a greek letter and a number.

 

[amb]

A bit of circuit description is as follows.

D1, Q1 and Q2 form a simple voltage regulator to provide the +/-12V rails for use by the circuit. U1, a BUF634, is used as a virtual ground rail splitter. This chip is used for its high current capability, and is needed because the relay coil current flows through it. VR1 is a trimpot that sets the zero reference voltage.

Q5 and Q6 are two transistors in a Darlington configuration for driving the relay. The time delay is accomplished by the charging of C7 through R11 after turn-on.

U2 is an inverting summing amplifier with a voltage gain of 10. It is also limited in frequency response with a low-pass corner frequency of 1.6Hz. This amplifies the DC offset signal from all the channels being monitored, which is then fed to Q3 and Q4. These two transistors provide positive and negative DC offset detection, respectively, and shunts C7 if the offset exceeds their turn on thresholds, causing the relay to disengage.

The schematic shows four inputs and four outputs (L+, L-, R-, R+) which corresponds to a four-channel (balanced output) amp configuration. For a three-channel amp, just use one of the "-" inputs and outputs for the shared ground channel. For a traditional two-channel amp, connect the "-" lines to the amp's ground.

 

[rickcr42]

Will we be seeing the pc boards offered in a "group buy" ?

because if yes be aware that other than present "buys" all the rest are being put on hold for now until this gets sorted out and an "easily understood" policy is put in place concerning selling in DIY

thanks

rickraptor

 

[BrokenEnglish]

that's great, amb! many thanks!
a simple question:
could one replace u2 with tl071?
thanks!

/edits: ah... you updated your first post in the meantime....

 

[n_maher]

:warpath:

amb,

Nice circuit, even if I can't understand it

.

Nate

 

[gsferrari]

[size=small]Can someone double check parts list for me:[/size]



Parts List:


Resistors:


R1, R2 = 3.32K
R3, R4, R5, R6, R7, R8 = 10K
R9 = 100K
R10 = 1K
R11 = 221K


Capacitors:


C1, C3, C4 = 22uF/25v Tantalum
C2 = 0.1uF Multi-layer Ceramic (NP0)
C5 = 1uF MKP
C6, C7 = 100uF/16V Non Polarized Electrolytic


Diodes:


D1 = 1N5252B
D2, D3, D4 = 1N4148


Semiconductor Devices:


Q1, Q3, Q4, Q5, Q6 = 2N3904
Q2, = 2N3906

U1 = BUF634P
U2 = TL081



Others:


Relay : Tyco RTE44012F / 12V dc / DPST-NO
VR1 : 1K ohm multi-turn cermet trimpot

Perfboard
Wire
I/O contacts

Additional Notes By amb
Code:

[left]1N914 can be substituted for the 1N4148 diodes, and the transistors are all non-critical (2N4401 can be used instead of 2N3904, 2N4403 for 2N3906, etc). For power supply voltages higher than +/-18V, put small TO-92 clip-on heat sinks on Q1 and Q2. The supply voltage should not be lower than +/-12V. Depending on ventilation you may also need a small heat sink on the BUF634P.[/left]

 

[dsavitsk]

Quote:

Originally Posted by gsferrari C5 = 1uF (cant make out the type!)

Looks like MKP to me. Wima perhaps?

redacted due to not caring anymore

 

[Garbz]

AMB I what happens at poweroff? Does the relay colapse fast enough to prevent poweroff thump too?
It's about bloody time someone made something like this!

Also Rick my Dynalo wouldn't be where it is without AMB, it happens to be his area of expertise eventhough not that many people build them.

 

[amb]

Quote:

Originally Posted by BrokenEnglish could one replace u2 with tl071?

Yes. Many different opamps could be used here, but I suggest FET input opamps only, and stay away from anything "cranky". There is no performance requirement here other than low intrinsic DC offset, so a jellybean TL081, TL071, LF351, etc., are all ok.

Quote:

Originally Posted by gsferrari Can someone double check parts list for me: ...

Yeah, your list looks fine. C2 should be a 0.1uF multilayer ceramic (NP0) cap, and C6 should be non-polarized electrolytic (also called "bi-polar electrolytic"). C7 doesn't need to be non-polar, but it's easier to just get the same type as C6.

VR1 is missing from your list, and it should be a 1K ohm multi-turn cermet trimpot.

1N914 can be substituted for the 1N4148 diodes, and the transistors are all non-critical (2N4401 can be used instead of 2N3904, 2N4403 for 2N3906, etc).

For power supply voltages higher than +/-18V, put small TO-92 clip-on heat sinks on Q1 and Q2. The supply voltage should not be lower than +/-12V. Depending on ventilation you may also need a small heat sink on the BUF634P.

 

[amb]

Quote:

Originally Posted by Garbz AMB I what happens at poweroff? Does the relay colapse fast enough to prevent poweroff thump too?

The relay will open as the voltage from the supply rails drop. Depending on how much rail capacitance you have on your rails there will be a delay, but hopefully the relay will disconnect the headphones before the amp begins to emit any last gasping noises.

One neat by-product of having its own virtual ground reference, this circuit will draw the same amount of current from each rail (the relay current is sourced from one rail, and sunk by the BUF634 and redirected to the opposite rail), so it will not cause an imbalance in the drain-down time of the positive and negative supply rail caps after turn-off.

 

[amb]

Quote:

Originally Posted by SHLim I for one would prefer a pcb and I don't want to spend time tracing my work to make sure the wiring is correct.

If someone wants to layout a board and do an order, be my guest. I have no financial interest in this, so do whatever you want with the schematic. As I said, it's my gift to the community, and if someone has good ideas to improving the circuit, I'm all ears!

 

[DaKi][er]

Quote:

Originally Posted by amb If someone wants to layout a board and do an order, be my guest. I have no financial interest in this, so do whatever you want with the schematic. As I said, it's my gift to the community, and if someone has good ideas to improving the circuit, I'm all ears!

it wouldnt be hard to make a perfboard template layout for this and just p2p it on that

just i dont have much of an interest in building one, so i'm not doing it

 

[BrokenEnglish]

amb... as far as i understand, i could connect the voltage-splitter part directly to the amp's psu jack and it hasn't to be a dual-psu, that's feeding the amp, am i right? could you please elaborate on how to adjust "zero reference". thanks!

oh... if only vksy could jump in to draw a nice p2p-layout as he did with the steps... things would be a lot easier...

...i see myself sticking part for part in a perfboard, arranging and rearranging it a whole weekend through...

 

[amb]

Quote:

Originally Posted by BrokenEnglish amb... as far as i understand, i could connect the voltage-splitter part directly to the amp's psu jack and it hasn't to be a dual-psu, that's feeding the amp, am i right? could you please elaborate on how to adjust "zero reference". thanks!

Yes, just take the V+ and V- rails from your amp to power this circuit. No need to connect the ground from your amp to it (whether it's a "real" or "virtual" ground). If you have a separate PSU for each amp channel, you should use a separate ε12 circuit per channel too. Do not use a separate PSU for this circuit, otherwise the zero reference cannot be correctly established.

The zero reference adjustment pot establishes what this circuit considers as "0 DC offset".

For a 2-channel amp (e.g., unbalanced dynalo/dynahi, SDS Labs, Eaton, etc.), just adjust the pot until the voltage between the amp's ground and pin 6 of the ε12's BUF634 is zero.

For a 3-channel amp (e.g., M³, PPA, Pimeta), adjust the pot until the voltage between the amp's ground channel's output and pin 6 of the ε12's BUF634 is zero.

For a 4-channel amp (amp with fully balanced outputs), this is a bit trickier. Since there are two active "ground" outputs, one per channel, each having a small DC offset during normal operation, you should adjust the pot until the voltage between pin 6 of the ε12's BUF634 and both of the amp's "ground" outputs are the lowest you could make them.

This assumes a correctly-functioning amp that does not have a problem with high DC offset. The ε12 is there to protect your headphones in the event of a malfunction.

 

[BrokenEnglish]

thanks for your answers, amb!
another question: since i have lots of ha-5002 on my hands (thanks to glassman!

), could they be used as replacements for u1 (after rerouting the pin-connections)?
thanks!