Dual star ground wiring in mono power supply

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi everyone,
I am in the process of building a headphone amplifier incorporated to a DAC.
The plan is to build it like the following lampizator project according to this schematic (link to his webpage) : Preamplifier_with_HEAD

I am going to make only a small modification to the power supply by using 3 tube rectifiers, 3 separate transformers and a choke input.

Since I know the interest of a sound ground design I draw some ways to physically build the thing... any advices are welcome, suggestions or questions.

An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.
 
Wow. I'm not really an expert but it looks like asking for trouble, to me, actually.

How about using star grounding?

EACH point to be connected to ground would have its own completely separate conductor that goes all the way back to the single star ground point, by itself. In practice, maybe some similar-signal-type ones could share a conductor. But things like grid input resistor ground reference points' ground-return conductors should never share any length of ground-return conductor with anything else.

Ground-return currents induce voltages across the conductlors' distributed impedances (current times resistance plus rate-of-change of current times inductance), which effectively show up back at the non-ground end of every connected conductor that shares any length of ground-return path with them. For example, for the grid-input ground reference points, that would cause the induced "ground bounce" voltages to be arithmetically summed with your grid input voltages. Not good.

Note, also, that the LOOPS shown in your schematic will also have currents induced in them by any time-varying magnetic or EM fields in the air (AC power, for example), in proportion to the enclosed loop area, and that those currents will induce voltages across the distributed impedance (mostly resistance and inductance) of the loop conductors. Not good. Similarly, a loop can radiate EM fields, too.

Regarding the star ground point: It's often located around the midpoint of the ground connections of the main smoothing caps in the power supply.
 
Last edited:
Thanks for the reply; I can see that the ground I made is not localized at the capacitor bank but just floating in the middle.
I revised my schematic and put 4 star grounds at capacitors C2, C 3, and C5, C7 decoupling their relative circuit power line. I will post it soon.
To answer your last question, the dac is simply the akm4396 differential output. The reason for connecting the V- to the cathode is to nullify noise by the difference of voltage variation from grid to cathode from one signal to another. Having said that, I can't follow what you mean by ground ends of grid input resistor... I didn’t put a grid input resistor, only a grid loading resistor. There is a cathode resistor of 200R and the V- is connected directly to the cathode so the voltage difference is 5.6 V (2.8V +2.8V) between the grid and the cathode. The cathode effective reference is 0v, there is no cathode resistor to ground to elevate the reference voltage cathode-grid. Thanks for pointing that out because I found a solution to this.
I realized that I won`t use the V- signal because the 200R is too low.
 
My last revision. The idea behind it is to use it both as a preamplifier and headphone amplifier. There is 3 separate power supply. One for the dac output and two others for the headphone left and right channels. Now do you see ground loops !? The second picture is an alternative way of grounding the output directly to the capacitor banks, or maybe i should use only of of the capacitor bank...


An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
You probably first need to look at pages 34-36 of the AK4396 datasheet at http://www.asahi-kasei.co.jp/akm/en/product/ak4396/ak4396_f00e.pdf . If that is the same chip then the differential outputs need to be low-pass filtered and also converted from differential to single-ended, before you use them. That will give you a signal and signal ground conductor for each channel.
 
Administrator
Joined 2004
Paid Member
Hi , thanks for pointing this to me. I thought using only the + signal would work, the - signal would not connected.

It will work, but substantially degrades the performance of the DAC. The benefit of balanced operation is that even harmonics and noise appear as common mode signals on the output cancel when combined.

This design seems needlessly complex, a high quality 1:1 transformer will do your balanced to unbalanced conversion, giving you great common mode rejection and a single amplifier stage after that ought to be all that is required.
 
I build it and of course it was quite a challenge... First, do not build the circuit as shown; ill remove it, because the headphone section has all the wrong values calculated by lampizator website. (srpp correct voltage 165V, Headphone section : Ra=7500 Rk1= 200, Rk2 = 6500 V=165)
There is some parasitic noise in the volume control metal casings and the signal wires from the dac to the headphone amplifier. One tube diode socket makes the ez81 noisy, I can’t figure why yet.

The biggest problem is the dac output going to the srpp tube. The dac signal seems over-amplified and the dac is clipping badly when music is louder. I will try to load the input with 10K instead of 240K or 75K (made no difference). I will also try to put a 1k resistor in serie to normalise the output impedance of the dac.
 
Project : completed !
I used a wolfson dac with build-in opamp, similar to tda1543
The circuit is simple, volume control shunt 100k input.
- cathode follower 1/2 tube left and right , 150uf output capacitor
CLCRC filter, AC heaters, EZ81 rectifier.
Sound ? Awesome ! no hum, no noise, perfect amplification, just a bit loud , I recommend lot of volume control attenuation !
 
cathodefollower%20headamplifier.jpg


Important Parts : Hammond 369EX, Choke Hammond 181R 20H, EZ81, 6n6p,
Specifications : tube bias 8.8 ma on 1/2 6n6p, grid current 0.0005ma 1/2 6n6p
Grid bias resistors : grid-HT : 355k, Grid-Ground : 145k.
Power rating : 0.1W , 300R Headphones recommended.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.