[Pars]

Thanks Amb. Yes, I have breadboarded everything except the power supply portion (don't have an OPA551 laying around at the moment). I had questioned the need for a couple of the diodes (D4 and D6), and although I don't think they're really needed I left them in. I home-etched a board last nite and will be prototyping it... may have a few commercially done also (if so will send you one). PM me if you have any concerns, etc.

As far as a choice of Schottky goes, I just picked something in a reasonable pkg. with the lowest Vf I could find and the BAT43 seemed to meet those criteria. I personally won't build mine that way though.

 

[Nisbeth]

Nice work, Chris

A little nitpicking though, just because I can't help myself

Your groundplane renders nicer if you:
- Put another angle in the trace running from LED2 to Q8 so that it just slinks around LED1
- Turn Q7 so that it faces the same way as Q1
- Put another angle in the trace from D5 to R12 (so that it runs closer to the groundpad on C6)

Also, I think you should move the bottom right mounting hole to the corner of the board. You might have to move LED2 a little but it's not much (and if you think in 3D it's probably not necessary at all, since the LED will be above the board surface)

/U.

 

[Pars]

Thanks Uffe. I'll take a look at it. I intended to leave plenty of space around the LEDs for some Lumex lightpipes...

 

[ottopig]

Okay, now that all of you who really know what you are doing have done all the heavy lifting and made it workable for us noobies, are there any plans for a group buy or another way to obtain an "e12" board besides etching or ordering our own?

I'm really interested, with an M-H and and M3 each at home and work, and four M-H waiting for cases and then to be given away. That's six. PM me if your answer is too "commercial" for the DIY section.......

ottopig

 

[Pars]

Quote:

Originally Posted by ottopig Okay, now that all of you who really know what you are doing have done all the heavy lifting and made it workable for us noobies, are there any plans for a group buy or another way to obtain an "e12" board besides etching or ordering our own? I'm really interested, with an M-H and and M3 each at home and work, and four M-H waiting for cases and then to be given away. That's six. PM me if your answer is too "commercial" for the DIY section....... ottopig

After I finalize the layout and build a prototype, I might consider having some boards run. The extras would be at cost plus shipping, so I don't think that would violate the "commercial" aspect of things. I'll let you know. Note that it will probably take me at least a month for the prototype as I don't have all the parts and the wife seems to have plans for any leftover money presently

 

[n_maher]

Any thoughts of Group Buys should be run past a mod before opening it up for public discussion. See Rick's post in the Dynalo thread.

HTH,

Nate

 

[Pars]

Work is progressing on the home-etched prototype. The board came out OK (first time doing a 2-sided board at home). I currently have the 555 timer/relay drive circuit functioning (works great!), with the mute switch and the two LED drivers. A few things I found were:

1) Didn't have enough space between the pot (VR-1) and the OPA551. Rectified in latest board layout.

2) I was using the R207/7 resistor layout in Eagle which provides 7.5 mm of space. This is pretty tight with anything other than small resistors such as generic metal film, Yageo, etc. My preferred BC 0.4W will barely fit. Don't even think about Dale RN60s. I am thinking of increasing the resistor package to R207/10 with 10mm spacing. This should allow RN60s. I figured most people would build these with whatever they have laying around, and RN60s are popular. Any thoughts? Board size would not increase.

3) Relay problems: I hadn't noticed that the Tyco/P&B relay originally used (RTE44012F or RTE24012F) has a different pinout than the Omron G6A or G5A-V2. The P&B has NC-C-NO whereas the Omron is C-NC-NO. Gut instinct tells me the Tyco/P&B pinout is more common. In skimming thru the Digikey catalog it doesn't appear that anyone other than Omron uses that pinout. Also, I noticed that Omron had some specific Telecom grade relays that are even smaller than the G5/G6 ones. I was under the impression that most audio manufacturers used Telecom grade for their reliability? At any rate, should I dispense with trying to accomodate more than one relay type, or if I continue with this, what relays should I try to support?

Thanks for any thoughts/recommendations.

Chris

 

[steinchen]

2)
my vote: increase the resistor package to R207/10, since the board size stays the same it adds versatility without a drawback

3)
I already bought Tycos, but Omrons are available in Germany in contrast to the Tycos. All other parts should be easy to source round the world, so I think it would be great if the board supported both pinouts to make part sourcing easy for DIYers throughout the world.

 

[amb]

Hi folks, just as a FYI, I am adopting guzzler's modifcations as version 1.2 of this circuit, and will be updating the schematic and parts list on my web site very soon.

 

[Dougigs]

That's a very impressive PCB design, and a nice demonstration of communal circuit-design work. I currently have two of my amps 'protected' with a single-transistor + relay delay/muting circuit I hacked together on protoboard, and this is definitely going to replace it. Over the course of this thread I've gone from yawning disinterest in yet another power-on-delay-muting circuit to a real interest in this one, which is very nicely tailored to the world of headphone amps.

One teeny concern: If I'm hooking this up to an amp that has a bipolar supply (rather than simple + and -), do you think I'd be better advised to power it from both rails (i.e. skip the virtual-ground circuit)? My concern being that throwing its current draw on the + rail but not the - could create some asymmetry... this would not be a problem if the amp's supply is hugely overspecified with plenty of extra VAs available on both rails, but could be a problem if it's anywhere close to its limit, in which case this could pull down the voltage on the + supply without doing same for the -.

 

[amb]

Quote:

Originally Posted by Dougigs One teeny concern: If I'm hooking this up to an amp that has a bipolar supply (rather than simple + and -), do you think I'd be better advised to power it from both rails (i.e. skip the virtual-ground circuit)? My concern being that throwing its current draw on the + rail but not the - could create some asymmetry...

The beauty of the virtual ground circuit is that it is symmetrical with respect to current draw on the rails. Since the circuit has no connection to "real" ground, all draw from one rail must go to the other rail. In this circuit, the relay would have caused more draw on the positive rail than the negative rail if it was powered the conventional way. But the virtual ground neatly solves that problem.

 

[Pars]

Quote:

Originally Posted by Dougigs That's a very impressive PCB design, and a nice demonstration of communal circuit-design work. I currently have two of my amps 'protected' with a single-transistor + relay delay/muting circuit I hacked together on protoboard, and this is definitely going to replace it. Over the course of this thread I've gone from yawning disinterest in yet another power-on-delay-muting circuit to a real interest in this one, which is very nicely tailored to the world of headphone amps. One teeny concern: If I'm hooking this up to an amp that has a bipolar supply (rather than simple + and -), do you think I'd be better advised to power it from both rails (i.e. skip the virtual-ground circuit)? My concern being that throwing its current draw on the + rail but not the - could create some asymmetry... this would not be a problem if the amp's supply is hugely overspecified with plenty of extra VAs available on both rails, but could be a problem if it's anywhere close to its limit, in which case this could pull down the voltage on the + supply without doing same for the -.

Not sure which layout you're referring to (Guzzler's or mine), but I'll reply anyhow

Regarding your concern when used with an actual dual-rail supply, this was my intended use also. My observations in building both a circuit on Guzzler's board and on the home-etched board that I am currently building is that I do see some pulldown on the +V side, at least on the muting board itself. I would be very surprised to see any pulldown on my Gilmore PSU from this, but I haven't tried that just yet. Please note that I do not have any OPA551s handy, so was just testing it with a TL071 in the splitter portion (which is inadequate, as had been noted previously). With the relay energized on my board, the rails on the muting board drop from nearly balanced at around +/- 11.6V to +8.6/-13.4 or so (this was on my board as I also don't have the small form factor Omron relay to finish testing Guzzler's board). Also, from my initial testing, it appears that it is not possible to balance badly mismatched rails via the pot, but the general idea with the pot is to match the board ground to the amp ground as closely as possible. I am using a bench supply setup for +/- feeds and just feeding it the +V/-V sides without tying its ground in (i.e., set for ~+/- 15.5Vdc and feeding it ~31Vdc between the rails). Perhaps its something wrong with my test setup.

Please correct me if I'm wrong on any of this Amb. The circuit in my testing on a breadboard really does work as advertised WRT offset detection and shutting things down. I am in the very early stages of testing, and obviously am still without some parts

 

[amb]

Pars, if you use a OPA551 you'll find that the problem you encounter to essentially disappear. The TL071 does not have low-enough an output impedance to serve as an effective virtual ground splitter that is being asked to sink many tens of milliamperes without some voltage shift.

 

[Dougigs]

Quote:

The beauty of the virtual ground circuit is that it is symmetrical with respect to current draw on the rails. Since the circuit has no connection to "real" ground, all draw from one rail must go to the other rail. In this circuit, the relay would have caused more draw on the positive rail than the negative rail if it was powered the conventional way. But the virtual ground neatly solves that problem.

Sorry, I think I wasn't clear enough. I am indeed now a believer in rail-splitter circuits (from my good experiences with the M3 and the MINT). I had been concerned with the possibility of attaching this thing to the + and ground rails of an amp that has + and ground and - rails -- its current draw with the relays on seemed to be enough to load one rail. I have some amps and DACs with true bipolar supplies that I'd want to do this with.

But my question is moot if it is indeed possible to hook this up across the + and - rails rather than the + and ground. The question is, is this circuit happy with 36 volts across it? The OPA551 has a stunning 60V supply capability, so maybe that's the case (and maybe you've discussed this already and I've missed it).

 

[amb]

Dougigs, this circuit is intended to be powered from the + and - rails of the amp. Its V- is not intended to be connected to the amp's ground, whether real or virtual.

As for voltage handling, yes, the OPA551 can handle lots of it, but even before that, the pair of 2N3904/3906 serves as a voltage regulator so the OPA551 never even sees any more than +/-12V.