[threepointone]

[EDIT]
New situation: I've got an Alien DAC done, I've got a bit of a deadline for myself on this project, and I'm not going to be able to build anything else serious. The offset on the outputs of the DAC, iirc, is on the order of about 20mV or so from the ideal virtual ground. I'd rather not put in output caps, since all I have are crappy ones and the good audiophile ones I have either aren't big enough for use with an amp or are too big to fit my current designated case. Some of the amps which will be used with the DAC don't really have positions for input caps.

The DAC and amp power supplies are completely isolated from each other, and come off of different transformers, so there's no potential between any of the input/outputs.

Here's my quick + dirty design:
Code:

[left]LEFT/RIGHT CHANNELS +5V | | | \ / DAC SIGNAL OUT ---*-->\ 10K TRIMPOT | / | \ | | *---------- SIGNAL OUT | | | ground VIRTUAL GROUND CHANNEL (now that I think of it, could also be just the usual TLE2426 config +5V | | | \ / 10K TRIMPOT \<-------- VGND / SIGNAL GROUND OUT / \ | | | | ground[/left]

(original post below)



I'm doing a bit of testing with a simple cmoy design. I really don't want to have to stick input/output caps on the amp, and I can't complicate the design with any more active components. Currently the virtual ground is made up of a TLE2426 railsplitter with 2 1uF caps at the output, from V+/V- to vgnd. If I put two trimpots, one from each rail to the TLE2426, would that allow me to adjust the dc offset / without any more serious complications? This is running off of a well-regulated power supply and a known source, so the offset from the components should be constant.

 

[nikongod]

chances are good that the tle2426 will put vastly overpower the trimpots.

in place of a cap, can you use a transformer somewhere?

why not a cap?

 

[threepointone]

I'm going to use this to test a couple of amp designs and opamps, some of which are bipolar. From what most people experienced with the PINT, the use of an input cap = lots of noise (or a couple volts of offset at the output, if that's any better).
I might try transformers, but I don't have any suitable ones at all. Either way, I'd prefer if I could do this without sticking something in the signal path.


what do you mean by power? I think the TLE is current limited to something like 35mA or so, and most trimpots are rated 1/2watt. I'm not entirely sure what the voltage drop across the trimpots are, but assuming the worst-case-scenario the voltage drop across one of them is the supply voltage, at 12V and 20mA current, the power is .24W. In reality, i'm pretty sure it's a lot better than that.

 

[Polaris111688]

Got a schematic of your design?

 

[nikongod]

its odd that you say that about an input cap. i dont doubt that people had issues with the pint. i suspect that the issues you mention stemmed from other areas of the circuit than the input cap.

an insufficiently "decoupled" psu often hisses. listen to a grado ra1 based on the schematics, then add in 2 470uf caps across the too-small tantalums. this is not to bash the pint, but to say that LOTS of stuff can cause a hiss.

the weird thing is that the meier designs all use lm6171/6172 opamps, and all use input caps with no issues to speak of. there is an amp that uses bipolar opamps (name escapes me, it was SPECFICALLY mentioned when the pint was initally introduced) but no input cap. the design made some compromises regarding what appeared to the opamp to be a variable load to ground on the input. the cap insures that the input of the opamp NEVER sees the pot as a dc "dump" and never gets the input offset currents all screwed up.

put a nice big low value cap in between the pot and the grounding resistor. build the PSU so that it will not be the limiting factor in your design.

sometimes less is more, sometimes its just too little

 

[threepointone]

*bump* New idea for offset above; I'm coming back to this question after working on the other parts of my project.

In the PINT, pretty big resistors had to be put in the feedback loop in order to balance the offset current from the bipolar opamps. From what I've heard, the noise is quite noticable when the input caps are in. The thing is, the pint needs small input caps for portability, so the input caps go right before the 100K resistor. the feedback loop ressitors must equal that 100K resistor to balance the input impedances, and you get lots of thermal noise since you'll end up using a 620k resistors along the line. I suspect that Meier's designs use the input caps before the pot, but then you'd need probably at least a 4.7uF cap which wouldn't fit very well in an altoids tin.

anyway, does the new design in first post make sense? Is there anything wrong with it, other than the fact that it's not very elegant and that it'll consume an extra 0.5mA idle?

 

[Xakepa]

Regular 10K trimpot might not be temp stable down to .5mV...they tend to drift much more than metallic film resistor

 

[threepointone]

i was using multiturn trimpots, so it would be.

but either way, let me just warn anyone trying this: doesn't work, does all sorts of weird things. Apparently it starts clipping (or sounded more like a really messed up low pass filter?) pretty badly if you try to trim it out from the DAC directly to headphones. I also tried connecting it to an amp, and the clipping disappeared, but lo and behold: I made myself a really nice AM radio! One of the local AM radio stations was quite literally way stronger than the music signal I was passing to the amp.

I'm just going to stick in some caps into the design and forget about it, but for the sake of curiosity, does anyone know why it doesn't work? I don't know how the inside of a DAC works, and it probably has to do with something in there.

 

[robzy]

The later parts of this thread deal with a couple of different suggestions on how to use an alien dac without output caps:

http://www.head-fi.org/forums/showthread.php?t=195341

Is was pretty much decided that it wasnt worth the time/money/effort for a DAC like the alien dac.

Rob.

 

[flecom]

i am not 100% sure if it would work but what you could do if you want to be really clever is figure out what the DC-offset is from the dac (assuming it dosent change) and then run the signal into an opamp and the same amount of offset (with no signal) into the other side of an opamp... and use the opamp to get rid of the offset...

if the aliendac had used a balanced output chip you could just run the inverting and non inverting outputs into an opamp and get rid of the offset that way also

 

[picklgreen]

If you have that much offset in your outputs than you need to fix your BROKEN alien dac. You should NOT have that much offset with or without CL/CR. Check that your voltages are correct. Also what other parts did you leave out?