AMB Laboratories DIY Audio

[skinner4251]

New guy here. Has anyone else built a PSU to power these units? I haven't seen any posts for it (unless I missed them) and was wondering if this was overkill compared to what others are doing with their A20s.

I was looking to build an A20 unit (haven't decided on the active ground just yet) and I was considering the Sig22 as a nice compliment. I have built a 15V Sig22 PSU, and have built 2 A20 boards for use as a headphone amp.

I say if it's worth being done, it's worth being overdone. A nice tracking PSU should eliminate most problems, and most companies cheap out on their power supplies. Are there any issues with doing this? If I get bored with this, I'll swap out the parts to convert it for a B22 PSU.

Comments are appreciated. Thanks.

[amb]

σ22 is the default PSU for the α10 preamp analog section, which uses the α20 modules. Have a look.

[skinner4251]

Then I suppose it's a good choice, although I won't be making a full stereo preamp with all the bells and whistles like the A10.

[skinner4251]

The circuit is essentially a center-tapped bridge rectifier driving two bank caps, two filtered supply rails driving two discrete op-amps, a reference, some feedback components, and two pass elements.

The bridge is going to contribute more noise on the zero-crossings than it should. Thank goodness for sinusoidal power or the thing would scream. There should be resistors in series with the diode shunt caps. The resistor and cap values need to be set empirically to correctly cope with the diode capacitance and circuit inductance to kill high-frequency ringing on diode turn-off. Using fast diodes makes the snubbers burn less energy thus reducing the size of the components.

Maybe I'll look into an op amp version with a precision rectifier and buffered stages, would take more time than I have available but It might be nice to provide another option. The newer TI low noise stuff seems pretty good, provided there is a large gain BW product... just some thoughts...

Has anyone modified the circuit to improve performance yet? Has anyone touched it much since 2006?

[amb]

Despite what you say, it works very, very well, and needed no "improvement".

Yes, the rectifier snubber caps "should" theoretically be tuned to the rectifiers and have series resistors, but that would make the power supply difficult to build for people who must use different rectifiers and have no way of doing such tuning. In practice I did not observe any diode induced noise at the power supply output. Depending on the output voltage setting (and gain of the error amp), the output noise is usually in the single-digit µV range.

[linux-works]

aren't the MUR series diodes 'soft' anyway?

with soft diodes, do you even need or want snubbers?

[cobretti]

Despite what you say, it works very, very well, and needed no "improvement".

Yes, the rectifier snubber caps "should" theoretically be tuned to the rectifiers and have series resistors, but that would make the power supply difficult to build for people who must use different rectifiers and have no way of doing such tuning. In practice I did not observe any diode induced noise at the power supply output. Depending on the output voltage setting (and gain of the error amp), the output noise is usually in the single-digit µV range.

Yep. I just measured Sig22 today and measured 4.59uV on positive rail. I have higher noise on negative rail and I don't know why. I will have to investigate the problem. As a precaution, I removed all diodes bypass caps as they really can make things worse, or not, if they are precisely calculated...

[skinner4251]

I meant no offense. I have all the necessary equipment and knowledge at my fingertips, and most DIY guys probably don't. Making an entire project easy to build is paramount. Trust me, I don't think it's a poor design Ti. Understanding the circuit's operation and design intent and making tweaks is what DIY is all about (for me). I also have the luxury of working a few feet from an engineer who has a doctorate in magnetics and work in a plant that builds xfrmrs (winding, dipping, baking) so I'm spoiled. I also have acess to use Altium for a layout and Solidworks for the enclosures. So, again, I'm spoiled. I agree that the PSU and amplifier sections are stable, and I'm planning on building and modifying a few more. The A20 is very versatile and, depending on the component selection, should be extremely stable. Again, I meant no offense. Plus, everything in the forum is a great wealth of knowledge about practical audio design.

[amb]

No offense taken at all. If you're interested in the σ22's development history, check out this headwize thread.

[cobretti]

Ti, I was reading little bit about Sigma22 on Headwize, the above link you posted. I also have LNMP. My LNMP has noise with open or shorted input of 0.04uV measured with Fluke 87 V, it takes a while but the Fluke meter really displays 0.04mVAC. 40nV, not a bad number. You mentioned in the thread your LNMP gets 15uV with shorted inputs. What op amps did you use? I first used AD797 but I got high freq. oscillation, so I changed them to OPA228. Did you check your LNMP for oscillation?