This circuit is now at revision 1.2, incorporating guzzler's modifications.

The official ε12 website has been updated to reflect the rev 1.2 changes.

Here is the latest schematic diagram:

Here is a picture of an assembled board designed by guzzler:

[Edited by amb on 02-02-2007 at 04:47 AM.]

This probably isn't an issue at all for headphone amps, but it is a concern if you want to use this board for power amplifiers (and it looks like a great unit for protecting power amps).

While exposing a headphone to 1VDC for .5sec won't damage it, exposing a tweeter to 30VDC for that long could indeed blow it up -- so you need fast relay response.

There are some who believe that the use of a standard relay-damping diode, as you're using, can slow the relay down too much. I'd point to the work by some other DIYers on a similar circuit -- read the part of this article toward the end, where the speed of relay response is discussed:

http://www.sound...m/project33.htm

I don't know how much this is really a concern for this unit, but it's worth considering.

---

quote:

[By dougigs]
This probably isn't an issue at all for headphone amps, but it is a concern if you want to use this board for power amplifiers (and it looks like a great unit for protecting power amps).

---

If one really wants to use this circuit for speaker protection duty, there are a few modifications necessary:
1. Use a relay with higher contact rating
2. Due to 1., the coil current of the relay is likely going to be substantially higher. Depending on the relay, the Q1 and Q2 transistors, as well as the OPA551 rail splitter may become inadequate. In such case the power supply to this circuit would need to be redesigned.
3. The gain of the opamp U2 should be reduced to avoid false triggering.
4. The zener diode in series with the existing damping diode as shown at the ESP site may be added if desired.

[Edited by amb on 10-14-2005 at 02:06 PM.]

Due to demand, I decided to order a small quantity of these boards, and they have just arrived today. I need to verify that they are good to use before making them available. Check the ε12 website for news.

Here is a peek at the board:

Top side:

Bottom side:

[Edited by amb on 10-21-2005 at 09:58 PM.]

This issue applies to the new rev. 1.2 board as well as guzzler's original.

Other than this, the rev. 1.2 board does not need to have its Q1 to Q4 transistors mounted in reverse, nor does it need the wire jumper fix.

---

quote:

[By teerawit] I have almost finished a BOM for the Epsilon 12. Anyone wanna double check my parts? :D

---

[Edited by Mister X on 06-20-2006 at 02:18 AM.]

I will also check MisterX's parts cuz his may be higher quality :)

Also, the asterisked PN you have on the Mouser list for the 22uf cap is actually a 2.2uf (225 vs. 226)

[Edited by Pars on 10-27-2005 at 08:56 AM.]

Pst... you can get cheapies from allelectronics dot com for $.75 each
#DT-2225

[Edited by Mister X on 11-08-2005 at 02:08 AM.]

T491D226K025AS (also on my spreadsheet)

Is this a good one?

---

quote:

[By teerawit] I found some Kemet tants at Digikey for 65 cents each.

T491D226K025AS (also on my spreadsheet)

Is this a good one?

---

Those are for surface-mount, not through-hole.

---

quote:

[By amb] Those are for surface-mount, not through-hole.

---

---

quote:

[By Pars] I was curious about the surface mount tants also... if they had sufficient span to reach the pads, they should work just fine, no?

---

in your website description, in the power wiring, I read the epsilon12 cannot be used with a single power supply 0-XV. But you stipulate it could be possible by modification to the epsilon12 circuit. Can you explain what modifications should be done.

It could be nice if this board could be used for millet or others too, without having to power the epsilon12 with another small wallwart psu.

Another question: With G1 input becomes G2 output, and vice versa, why this number crossing?

And a third one, ok, I understand this one will not connect immediately the headphone when power on, but what about power off. Instabilities and noise occur also at that moment. Is the epsilon12 quick enough to disconnect the headphones just before the circuit becomes instable after powering off?

Many thanks,

GregVDS

---

quote:

[By GregVDS] in your website description, in the power wiring, I read the epsilon12 cannot be used with a single power supply 0-XV. But you stipulate it could be possible by modification to the epsilon12 circuit. Can you explain what modifications should be done.

---

Yes, I am considering several options for this, and will post details when they are finalized.

---

quote:

Another question: With G1 input becomes G2 output, and vice versa, why this number crossing?

---

Yes, it's due to a silkscreen labeling error on the PCB (or you could consider it a cross-wired trace, either way). So, I simply documented it as such.

---

quote:

I understand this one will not connect immediately the headphone when power on, but what about power off. Instabilities and noise occur also at that moment. Is the epsilon12 quick enough to disconnect the headphones just before the circuit becomes instable after powering off?

---

It should. The actual delay depends on how quickly the power supply rails drop in voltage after you turn the power off. When the voltage at the relay drops below a certain threshold it will open its contacts, and this should happen before any noise is emitted on most amps.

[Edited by amb on 11-06-2005 at 06:26 PM.]

This type of amp almost always have its V- as the "real ground", and have an output coupling capacitor because the nature of such amps is that the output will be biased to one-half the power supply voltage.

After thinking about this a bit, I have come to the conclusion that with this type of amp, there is a real need for turn-on and turn off muting, this is due to the charging and discharging of that output coupling cap. However, there is very little reason to use the DC offset protection portion of the ε12 because the output coupling cap will block the DC. The likelihood of that capacitor failing and leaking DC is quite slim.

Given this line of reasoning, the solution becomes quite simple -- retain the muting delay portion of the ε12, and skip the DC offset sensing part. This removes the need for a negative supply rail in the ε12, and hence also eliminates the need for the virtual ground rail splitter on the ε12 as well.

So, this means the following:
1. No need to populate the following parts:
U1, U2, R2 through R10, VR1, Q2 through Q4, C4 through C6, D3, D4.
2. Change D1 to a 12V zener diode (1N5242B)
3. Install a wire jumper in place of R2.
4. Install a wire jumper across the C-E pads of Q2.
5. Install a wire jumper in place of C4.
6. Use the amp's power supply to power the ε12 (12V-30V DC).
7. Connect the amp's left and right outputs to ε12's IL and IR.
8. Connect the ε12's OL and OR to the headphone jack.
9. Connect the headphone jack's ground contact directly to the amp ground (no need to use the G1 or G2 pads on the ε12).

This is absolutely the easiest solution, albeit with a small loss of functionality. To retain the DC offset sensing function would either require a separate and isolated power supply for the ε12, or some fairly hacky and complex changes and additions which I hesistate to recommend (because of additional parts needed that will not fit on the board).

What do you think?

[Edited by amb on 11-07-2005 at 06:17 AM.]