no on/off switch
→ Already fixed, then again, the switch is more part of the case than off the circuit board
the start-up delay will likely never start (because the resistors are both 47k Ohm)
the input stage Op-Amp needs some resistors before it to prevent external “whoopsies” from affecting it too much. Not sure if I add some.
the feedback circuit for the input-stage is complete bollocks
For the NE5532 I plan to use as test-dummy in the input stage, the datasheet specifies 36k Ohm resistors in the feedback circuit, recommends below 6k Ohm to optimize for noise.
For the fancy OP275 I plan to replace it with for use, the spec sheet says this:
In other words, Source resistance = Ground resistance || Feedback resistance is less than 2k Ohm. Schematic says… a little over 20 times that.
Cascaded Linear regulators
Should look like so:
Features:
start-up delay and DC-protection (yet to be tested)
Accepts DC from 16-ish Volts to 35-ish Volts. Not sure about AC. It does definitely not care about polarity of the power supply jack!
↑ Since exporting the above schematic, some more tweaks happened. One is to remove the whoopsie, the other is two include a pair of inductors on the power input
After this is built, tested and tweaked, I will post a final schematic.
Started work on the proto-board. Got the mount points for the potentiometer and one of the 6.3mm jacks carved out.
The Poti I got has 2.54mm grid size, which is great since the proto-board I have is the same. Except the “stability”-tongues, wich are just random. WTF Alps!? Why not 2.5 or 5mm spacing? WHY 3.3mm?!
To be fair, not a big deal. Just one of the many Dafuq?-moments in this project.
The 6.3mm jacks don’t claim to be for 2.54mm, so no surprises there.
I also found the heatsinks that went missing, too bad I ordered new (and different ones) in the meantime. These will be soldered into the board with the regulators screwed to it for stability (can’t have a TO-220 package flapping in the breeze, can we?)
Top-Side layout and routing mostly done. The only questionable decision is the black resistor on the right jumping over Gnd and Pwr for the input Op-Amp.
Completly missing are paths for audio
Sort of. I went with 6.3mm TRS on the input because I have spare cables to hook it up with.
And since it is unbalanced anyway, this saves me from having to drill one hole in the back panel. (As in 2 now VS 3 had I gone with RCA)
Also: $10 per piece? Seriously?
I could buy 28 of the jacks I used here (Neutrik NR-J6HF) for that money.
I have been thinking about putting in a notch filter to specifically filter out 1 kHz just to fuck with everyone trying to measure it.
One small notch in the spectrum missing should be mostly inaudible, but measurements are just worthless
Nah, that would be to detect extended 1 kHz and then putting a 1 kHz tuned oscillator on the output (effectively removing any harmonics).
Also considered doing that…
Should I ever bring something to market, be assured I will do my best to make measurement peoples lives miserable.
Blessed be thermal cut out!
In other words: No, skipping the bread-board prototype does not save time.
The longer story:
I wanted a virtual ground to circumvent me running after my own tail later with a ground loop causing mayham. The theory is solid. Have a positive rail and Negative rail, cut in half for virtual Ground.
As power draw varies depending on load, the resistors providing vGround need to vary. Easiest way to do that is with a pair of linear regulators.
My thinking here was instead of creating a positive rail before the vGround-circuit, I could do that after it. I was WRONG!
My shortcutted vGround is a very efficient 24W heater. I found that out by burning myself on a rather unhappy Linear Regulator in thermal-cutout (happens at >150°C / 300°F). Whoops!
The Fix:
Easiest way would be to bypass the vGround circuit. That would open the gates for ground loop issues and the fun that entails.
The “hackjob” would be to rewire the LM317 (= adjustable regulator) to be before the vGround circuit. This hack holds the possibility of not working.
Optimal solution is to man up and build the real deal, voltage reference, protection diodes, etc. The thing I wanted to avoid…
As this was meant as an easy starter, I will probably just kick the vGround circuit off the board and hope for the best.
20 Solder joints, 5 internal bits, 10 screws and drilling 5 holes away from seeing if this works
A few random questions on the schematic from Update 5. Some of it might just be because the schematic is outdated.
Where does the NE5532 power come from? I didn’t see the connections for it in the schematic (maybe I missed something). The NE5532 requires a positive and negative supply unlike the LM380N.
The LM380N’s output voltage is supposed to be centered at 1/2 of the supply voltage but the output connector common terminal is connected to GND. In the datasheet, the example circuits use a coupling capacitor to get rid of the DC offset. How does it work in this circuit?
The LM380’s voltage gain is 50 which seems like it might be too much (considering that the NE5532 also provides some gain).
Don’t know what happened to the rail labels. When done and working, I will update accordingly.
That is to be tested. All of this is a test bed.
In case it does not work, I have a selection of caps on hand to use.
NE5532 is unity gain stable, so I could configure that. I also have the option to bodge in a voltage divider at the InputL and InputR resistors.
As I document this project on two different forums, I may have missed the note of the NE5532 being placeholder for an OP275. I should also have another NJM4580 somewhere.
Note on split-rail: That would not be a problem (or a smaller one) if my oversimplified vGround would have worked.
To make this worse, before this project I had no need for multi-rail power, as such, my bench supply is single rail (and tops out at 15V-ish). So no bread-board sanity check either.
The Input voltage divider as I implemented it is just a pin-header to let me hook various resistors between the 20k Ohm and Ground. As drawn, it is NOT what it says on the schematic!
The “Pop-Surpressor” may need different resistors
The Label on the regulator providing the Aux-Rail is Bogus
You can totaly “Oopsie” an astronomical gain on the input stage
It is on hold until I figure out some important aspects I was not aware could be a problem.
The 5V Rail used to power the DC-detection circuit (marked obsolete), now it just feeds through a start up delay into the relay.
Don’t need to switch off on DC when there are capacitors to only couple AC
Those two resistors are there to get the output facing side of the capacitors into a known state. Else they are “floating” with what ever voltage happens to be in there when you unplugged the headphone.
Completely understand, sometimes is better to take a step back to a simpler thing before going to a more complex territory.
I see, it makes sense with the relay having the startup delay. My only concern would be if these resistors are somehow messing with the output impedance of the amplifier, maybe getting part of the power transfer to them. Although, considering how headphones are usually within the 100Ohm range, the hundred times bigger resistor should not make a significant difference.
Did you simulate this circuit? If so, what software are you using?
The resistors are in parallel with the headphone. As they are 10k Ohm and most headphones are 600 Ohm or less (= have 16 times less resistance), I am not expecting them to make a difference.
Kinda sorta? I am more hands-on when tinkering with electronics.
For simulations, I use LTspice. I am definitely not qualified to explain how to use it though.
)
