[walangalam]

Hi amb,

Just 1 more Question before I build this...

What happens when one channel is at -70mv and the other is at +70mv? Since its a summing amp, I would assume that the output of the opamp would be at 0mv, or is it something else?

Thanks

 

[amb]

walangalam, yes, the scenario you describe will null out at the summing amp. Keep in mind, though, that it is assumed that the main amp does not have such wacky DC offset issues (i.e., this protection circuit is not intended to mask an existing DC offset problem. The amp should normally have near zero DC offset). This circuit is there to protect the headphones in case the amp fails and the DC offset goes haywire. It is highly unlikely that two channels will fail at the same time but the offset goes in opposite directions with the same magnitude.

 

[amb]

By the way, this circuit now has its own web page.

 

[Garbz]

Just checking if the 1N4749 24V 1W zener is a suitable substitute for the 1N5252b?

 

[amb]

Quote:

Originally Posted by Garbz Just checking if the 1N4749 24V 1W zener is a suitable substitute for the 1N5252b?

Basically any 24V zener will work. The 1N5252B is rated at 500mW, the 1N4749 is overkill but will work.

 

[Garbz]

Overkill is only overkill if it's not the only part I can find

 

[Garbz]

Hey AMB is it possible to modify the design so the DC Epsilon2 can run from a single ended 24v supply? Such is the case with the amplifier where I was hoping to use this circuit.

 

[amb]

Quote:

Originally Posted by Garbz Hey AMB is it possible to modify the design so the DC Epsilon2 can run from a single ended 24v supply? Such is the case with the amplifier where I was hoping to use this circuit.

What amp are you considering? Does it use a rail splitter to get its virtual ground or is it actually single supply with an output coupling cap (i.e., ground is V-)? If it's the latter, the coupling cap will prevent any DC offset from reaching your headphones (unless the cap fails). If the cap is to fail then the DC offset can only go positive relative to ground, not negative. For such an amp, skip the BUF634 on the ε12, omit Q2 and Q4, connect all three pins where Q2 was together, and tie VG to the V- rail. Put a small heat sink on Q1 because it will be dropping 24V down to 12V and dissipating some heat.

 

[Garbz]

Yeah well the main problem is the coupling cap still produces some offset while it's charging and while the tube warms up. It's is fully single ended with -ve = GND and the output sitting at +12v or something I believe, I'll get exact measurments later.

The other problem is at poweroff the amp goes DC in a huge way. We are talking 1.2V into my dummy load. I'm hoping the relay will solve this issue.

 

[amb]

Yep, single supply amps with a big output coupling cap will thump big-time upon turn-on. The muting delay should take care of that.

 

[walangalam]

In my dynalo, one channel has a positive offset drift, the other has negative. At start up, its around 10mv either way. Without the servo, it climbs to around 35mv -- one channel postive, the other negative. But with the servo, its a cool 0mv.

The servo is a opa2132 serving both channels. If this chip fails, what I describe could happen, specially since my amp runs very hot.

So I plan double up the sense amp, use maybe a tl082 dual, and duplicate everything for U2 (use only 2 10K inputs) including R10 and sum at the base of Q3. Would this work alright?

I also found a more sensitive relay in digikey 255-1404-5-ND. Its rated at 4.2ma 12V, so i could use TLE splitter after all.

 

[amb]

Quote:

Originally Posted by walangalam In my dynalo, one channel has a positive offset drift, the other has negative. At start up, its around 10mv either way. Without the servo, it climbs to around 35mv -- one channel postive, the other negative. But with the servo, its a cool 0mv. The servo is a opa2132 serving both channels. If this chip fails, what I describe could happen, specially since my amp runs very hot.

If the servo opamp is to fail, chances are its output will drift toward one of the rails, causing a big DC offset at the output rather than in a way like you describe.

Quote:

So I plan double up the sense amp, use maybe a tl082 dual, and duplicate everything for U2 (use only 2 10K inputs) including R10 and sum at the base of Q3. Would this work alright?

No, because the two offsets will still sum at that point and cancel out. If you want completely independent monitoring of the two channels (which I don't think is necessary), you should use a completely separate ε12 circuit per channel.

Quote:

I also found a more sensitive relay in digikey 255-1404-5-ND. Its rated at 4.2ma 12V, so i could use TLE splitter after all.

That relay has a low current coil, but its contacts are only rated for 10mA (compared to 8A like the Tyco that I specified). 10mA is really too low...

Also, using a TLE2426 as the splitter will not allow you to adjust the zero reference voltage via a pot.

 

[walangalam]

Thanks. I didnt see the 10ma rating.

I did realize that it will still sum at Q3, but only after I had already posted! I think I will double up the e3. Now to find a way to use only 1 relay...

 

[Garbz]

you should really have the dynalo running at 0v without servo after it's warm, and let the servo just handle the amp heating up.

I.e. remove them power the unit up and prehapse adjust the 500ohm resistor feeding the LEDs to null the offset when the unit is on. That way if a servo fails you're safe.

Still it's unlikely both servo's will fail synchronously.

 

[BrokenEnglish]

could someone please post some pics of its build?! most of the parts are on their way and i'd like to decide on my parts layout or... if i build it at all. thanks!