So the shields of those cables will be antennas and together with the cable from your red circled point you have a dipole with your groundplane in the middle. You really want that red circled wire going as close to the 0V of those jacks as you can possibly get it - at present those RF currents are shooting past the opamp in the middle inducing all sorts of audio nastiness...
Ah but I can't actually see the 0V connections to those jacks - perhaps you have dedicated wires for them?
Ah but I can't actually see the 0V connections to those jacks - perhaps you have dedicated wires for them?
Ahh thanks I hadn't thought of that. I just moved the central ground point right between the jacks. hopefully once I'm done with this the filtering on the lines coming in will deal with the majority of what's riding on the input and output cables but I definitely don't want to have what's left traversing the ground path for the rest of the components.
The ground point for the jacks are the top most pin at the edge of the board. I still need to break that trace at the pins and put in the pads for the ferrites
If you fix this up you'll be ahead of most (practically all) of the commercial manufacturers I've seen. Its the single biggest SQ problem I reckon in commercial gear - having a dirty groundplane contaminated by CM currents.
The worst part is I'm doing all this work on a test board I made just to solve my RF issues. Once I have the LTE noise licked I have to carry over what I've learned to the real layout designed to fit in those metal cases a few posts back. I can't just toss jacks and stuff wherever I please. they have to fit the case ;(.
But once I'm solid on how to go about what needs done to clear up the signal bleed it's only a couple more hours to update the other board layout.
One thing is for sure.. I don't want to touch my power and charging circuits. Lets call it accidental skill
but I've plugged the amp into some really cruddy USB chargers and it's dead silent.. no whine or noise at all coming from the power delivery section.The DC/DC converter is operating at 2mhz.. if it's a bit dirty I wouldn't know it from listening since the noise would be far above the audible spectrum. I'll have to plug the amp into a scope once I've fixed the RF noise and see if I have anything left over coming from the DC/DC converter around the 2mhz mark.
Fixing the noise coming out of the DC/DC should be a walk in the park compared to reducing airborne RF effects The SQ effects won't be noticeable directly but indirectly if the 2MHz is bleeding into your opamps that'll reduce the dynamics and 'grey out' the tonal colours of your sound. Ferrites plus inductors will probably do the trick for filtering it.
The SQ effects won't be noticeable directly but indirectly if the 2MHz is bleeding into your opamps that'll reduce the dynamics and 'grey out' the tonal colours of your sound. Ferrites plus inductors will probably do the trick for filtering it.I have to say. For a single stage output I'm quite impressed by the sound of this little 4556AD op amp. I wanted to try and use something that had enough output to pull double duty as gain and output to save on layout space and I'm not disappointed.
My HTC rezound is pretty lame on output power. It doesn't even manage 0.8v RMS into 32ohm load before distortion goes through the roof. I've taken the 4556AD 30% past it's max rated output current per channel and it still sounds clean.
I have a few AD8397ARDZ on the way to retrofit into the amp and check them out. The extra output headroom will be welcome but I'm hoping they don't change the sound.
My HTC rezound is pretty lame on output power. It doesn't even manage 0.8v RMS into 32ohm load before distortion goes through the roof. I've taken the 4556AD 30% past it's max rated output current per channel and it still sounds clean.
I have a few AD8397ARDZ on the way to retrofit into the amp and check them out. The extra output headroom will be welcome but I'm hoping they don't change the sound.
One hiccup - the AD8397ARDZ doesn't have output current limiting like the NJM4556A does. The shorting that occurs when lugging and unplugging the headphone TRS jack eventually fries it unless the volume is always turned down, or some resistance is in series with the output. AMB uses 6.2R in series with the chip in the mini^3:
http://www.amb.org/audio/mini3/
There is probably a more elegant current limiting solution out there somewhere.
http://www.amb.org/audio/mini3/
There is probably a more elegant current limiting solution out there somewhere.
Last edited:
There is probably a more elegant current limiting solution out there somewhere.
Use a non-shorting connector. If you want to keep the 3.5 stereo jack on your phones, make an inline adapter to, say, micro XLR, and always unplug the XLR instead of the jack.
One hiccup - the AD8397ARDZ doesn't have output current limiting like the NJM4556A does. The shorting that occurs when lugging and unplugging the headphone TRS jack eventually fries it unless the volume is always turned down, or some resistance is in series with the output. AMB uses 6.2R in series with the chip in the mini^3:
The Mini³ Portable Stereo Headphone Amplifier
There is probably a more elegant current limiting solution out there somewhere.
I'll probably just do the same thing. I have more than enough voltage headroom (+/- 12v) to compensate for the loss.
I've heard of people using small filament bulbs to limit current on the output. The idea is the filament offers very little resistance when cold and passing normal current to the load. If the load is shorted or starts pulling excess current for some other reason the filament heats up and increases it's resistance.
You keep the excellent damping of a near 0 ohm output resistance and at least some sort of protection against extended shorts.
You keep the excellent damping of a near 0 ohm output resistance and at least some sort of protection against extended shorts.
It's not a question of lost voltage. You can put the resistor outside the feedback loop, in which case it raises the output impedance of the amplifier, which is ~OK for higher impedance phones, or you can put it inside the feedback loop, where it has some undesirable effects but principally with a reactive load (like headphones) the voltage drop across the resistor will be out of phase with the output of the AD8397.
Of course how serious these effects are is dependent on the size of the resistor and what significance you can ascribe to them is arguable perhaps, but it is surely better to simply avoid the potential problems if possible.
BTW, you can do a LOT better with that layout. It's partly cosmetic; you shouldn't have tracks that are dragged off the horizontal or vertical, and you shouldn't make rightangle bends, rightangles at T-junctions are unavoidable, but tracks should otherwise be chamfered and no included angle should be less than 90 degrees. This makes no difference at audio frequencies, but you should aim to develop good habits. When you produce a design that other people are going to look at, it's a good idea to orient the component labels so that they are easily readable.
I know that you are attempting to produce a prototype that will allow you to make modifications, but you have used about 4 times the real estate I would have used for the same design. You have only a few components, you don't have to worry about pick-and-place clearances, you can shrink the whole thing a lot.
I know that you are attempting to produce a prototype that will allow you to make modifications, but you have used about 4 times the real estate I would have used for the same design. You have only a few components, you don't have to worry about pick-and-place clearances, you can shrink the whole thing a lot.
The soft on/off power management controller I'm using has built in short circuit protection. It switches the DC/DC converter on and off. The DC/DC converter doesn't have any short protection itself but I can insert a current limiter between the power mosfets connected to the power management controller and then have the power management controller monitor the current regulator's voltage output. If it drops below a certain voltage then it will assume there is a short and start pulsing the output (turning the power mosfets on and off) till the condition is fixed.
I could just use a current limiter on the DC/DC converter but that thing will operate down to 0.5v and still provide 24v output. I'm not sure if any damage could occur if the current limiter were to try and feed it low voltages with high output loads.
There is probably a more elegant current limiting solution out there somewhere.
Just off the top of my head - my laptop knows when there's a jack plugged in. So it must be possible to sense when no jack's plugged in and disable the output - perhaps by turning down the regulators to the output opamp to their min voltage (1.25V typically). A time delay is presumably called for to allow for people who plug jacks in slowly or fiddle with them...
Just off the top of my head - my laptop knows when there's a jack plugged in. So it must be possible to sense when no jack's plugged in and disable the output - perhaps by turning down the regulators to the output opamp to their min voltage (1.25V typically). A time delay is presumably called for to allow for people who plug jacks in slowly or fiddle with them...
Most laptops use a jack with a small switch that opens or closes when a jack is fully inserted. I could use something like that to cut the power to the converter or power the amp down if no headphone plug is inserted. Depending on how far the jack pulls out before the switch triggers it could possibly still have some distance on the way out/in where it shorts but still in far enough to hit the switch. It's definitely something to consider and they make jacks with switched pins the same style I'm using currently in the amp. like 10 cents more per which isn't a big deal.
Some computer sound cards also take a guess at what type of device is plugged in and automatically configure the jack for that. I think they test the resistance of the device plugged in and maybe under 1k or so just assume it's a headphone or something to that effect.
Most people just settle for a TPA6120A or a pair of LT1206s, either of which are good for continuous short circuit.
The AD8397's principal claim to fame is that it will swing pretty near rail-to-rail, but it's only good for +/-12V. The other thing is that it's 2 opamps in a small 8-pin package, but given the problems with short-circuit, if you can't live with the XLR or some other way of avoiding shorting it, you'll just end up bolting on a load more gubbins to protect it, which will occupy as much space as using the other 2 chips I've mentioned.
The AD8397's principal claim to fame is that it will swing pretty near rail-to-rail, but it's only good for +/-12V. The other thing is that it's 2 opamps in a small 8-pin package, but given the problems with short-circuit, if you can't live with the XLR or some other way of avoiding shorting it, you'll just end up bolting on a load more gubbins to protect it, which will occupy as much space as using the other 2 chips I've mentioned.
Plus the added cost of the protection components.
I thought about it a bit. I was originally planning to current limit the DC/DC converter and feed the output voltage back to the on/off controller to power the amp off if voltage dipped too low but I think I might just move the current limiter to the output of the voltage source selector and battery charger IC. The on/off push-button controller has a voltage supervisor which will cut off the power under 3.1v anyway.
Havn't finished reading it all yet but some ground comments I have to disagree with.
Having contiguous ground planes top and bottom of a board, with multiple stitching vias, and vias round the edge of the board is one of the best methods for ensuring RF immunity.
Silly spider leg star grounding is hopeless and looks like a fractal antenna to the RF.
Having contiguous ground planes top and bottom of a board, with multiple stitching vias, and vias round the edge of the board is one of the best methods for ensuring RF immunity.
Silly spider leg star grounding is hopeless and looks like a fractal antenna to the RF.
Yeah I never actually built the star grounded layout. I've since moved on to attempting to find the proper components to filter the inputs and outputs as best possible before I move back to tinkering with ground planes and fill. I can't keep the cabling external to the amp from acting like antennas so I need to make sure I have that all cleaned up before I can tell what if anything layout changes are doing.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.