Single diode supply for DAC

[fedde]

Hi guys,

I was wondering if anyone of you has done some experiments with a single diode supply instead of a bridge with four diodes. I've heard that the Shigaraki and Progression DACs of 47 Labs use only a single diode.

I did some experiments and found out that there quite a difference in sound between these setups. I did not reach a final conclusion, but the sound with a single diode seems to be more detailed and open.

I think that the single diode supply could have more advantages beside cost, size and circuit complexity reduction. The diodes have switching noise and are a non-linear component, so I'd guess it is an advantage to have only one diode in the power supply path. A disadvantage is the greater ripple. This can be compensated for with a larger cap. Or you could see it as a trade-off between ripple and HF noise. The regulator after the diodes will have far less trouble with 50-60 Hz ripple than HF noise...

Any thoughts???
(or better: measurements with spectrum analysers 🙂)

Fedde

 

[ThorstenL]

Konnichiwa,

I was wondering if anyone of you has done some experiments with a single diode supply instead of a bridge with four diodes.

I only ever compared bridges to full wave rectification, generally preferring the latter.

I've heard that the Shigaraki and Progression DACs of 47 Labs use only a single diode.

So have I.

I did some experiments and found out that there quite a difference in sound between these setups.

I would expect so.

I did not reach a final conclusion, but the sound with a single diode seems to be more detailed and open.

Maybe just "brighter"? With the natural HF rolloff of the NOS DAC this may help ofsetting some of the "darkness"?

I think that the single diode supply could have more advantages beside cost, size and circuit complexity reduction. The diodes have switching noise and are a non-linear component, so I'd guess it is an advantage to have only one diode in the power supply path.

Maybe. But you also end up putting a significant DC load on the mains transformer. You must severely derate any mains transformer used with such a rectifier. Not an issue if you use a 170VA Transformer to draw 100mA or so, but potentially an issue if you use a "wallwart".

A disadvantage is the greater ripple. This can be compensated for with a larger cap.

I would suggest against this. Keep the cap's small(ish). Modern Regulators do well enough in killing ripple.

Any thoughts???

I suggest you compare single diode and full-wave rectification (needs centertapped secondary) to bridge. It might be interresting. If you find that Bridge and Fullwave sound similar or identical the cause for the change in sound is the higher noise level with the single diode or possibly the fact that ripple is 50Hz not 100Hz. If you find the Single Diode and Full Wave to sound similar then the HF noise signature is most likely the issue.

Sayonara

 

[fedde]

Thanks Kuei for your response!

So you suggest full wave rectification with two diodes !? Doesn't that also have the DC loading problem !? I do have a centre-tapped toroid, so I can try that... (well, a double secondary toroid, but that'll work too)

Brighter? Yeah, I guess it is brighter! Too bright for my taste, I have to say... (but I also use different components 4xMUR860 vs 1xSchottky, I'll do better experiments lateron)

"I would suggest against this. Keep the cap's small(ish). Modern Regulators do well enough in killing ripple."

Yes, that was the next experiment I wanted to carry out. I currently use 1000 uF P. FC, but that could be too large...

Fedde

 

[tiroth]

I would not use smaller caps than that! I would not expect more than 60dB ripple attenuation, so to put ripple below the noise floor of a good reg you still need to deliver < 5-20mV of ripple to the input of the regulator.

I was experimenting recently with 2200uF + 10mH/3R + 2200uF + R + 680uF next to the regulator. I would choose R to control the ripple, as the current draw was relatively constant. This would not work well for very dynamic current draws of course.

 

[moamps]

Re: Re: Single diode supply for DAC

I would suggest against this. Keep the cap's small(ish). Modern Regulators do well enough in killing ripple.

Hi,

With small caps we need higher secondary voltage for proper regulators work. Current impulse thru diode is than higher. HF noise is function of diode current. I see circle here.
Using two diode-double secondary full wave rectifier is IMHO good idea, half wave rectifying isn't at all.

Regards

 

[ergo]

One aspect in all this subject that I have been thinking of is that the 100Hz ripple has harmonics also.

So, you do not only get 100Hz + any HF garbage but also some harmonics @ 200, 300, 400 etc. Hz.

What I'm getting at is that perhaps there is one more point why half wave rectification sounds better. There you get 50Hz ripple + 100, 150, 200 Hz etc. harmonics + less HF garbage. All this means less colouration to the critical midrange, doesn't it ?

Ergo

 

[peranders]

If you have the opportunity, use a fullwave bridge. This is the cleanest way to rectify. Other methods creates more harmonics and losses in the transformer. It's nicer to the smoothing caps also with fullwave rectification.

 

[peranders]

Re: Re: Re: Single diode supply for DAC

Using two diode-double secondary full wave rectifier is IMHO good idea, half wave rectifying isn't at all.

Using a fullwave bridge is even better because using two diodes is also halfway rectification of each half in the transformer, which creates more losses and harmonics.

 

[moamps]

Re: Re: Re: Re: Single diode supply for DAC

Using a fullwave bridge is even better because using two diodes is also halfway rectification of each half in the transformer, which creates more losses and harmonics.

Hi,

Sorry, I don't think so.
Transformer is full wave loaded. Current flow thru only one diode all the time instead of 4diode-bridge where are two diodes active.

Regards

 

[peranders]

Re: Re: Re: Re: Re: Single diode supply for DAC

Sorry, I don't think so.
Transformer is full wave loaded. Current flow thru only one diode all the time instead of 4diode-bridge where are two diodes active.

Think how currents flow. Through each winding you will have current on a half period only. 50% of the time no current will flow. The transformer is driven harder with a 2-diode rectification. The thing is halfwave rectification times two with the two sections 180 deg separerated apart. This creates fullwave rectification with only two diodes.

 

[moamps]

Re: Re: Re: Re: Re: Re: Single diode supply for DAC

Think how currents flow. Through each winding you will have current on a half period only.

Yes

50% of the time no current will flow.

In ONE secondary. On primary side transformer see fullwave load.

The transformer is driven harder with a 2-diode rectification.

No. Same. Only secondary windings are doubled.

The thing is halfwave rectification times two with the two sections 180 deg separerated apart. This creates fullwave rectification with only two diodes. [/B]

Yes.
Two diode instead of four makes smaller noise. IMHO

Regards

 

[peranders]

It's the secondary windings which takes abuse (more losses). I gather also that the iron core is not used at optimum. 2-diode solution with centertap is a budget solution but in most cases I think the extra losses are of minor importance.

 

[Pedja]

2-diode solution with centertap is a budget solution...

No, it’s just the opposite.

Traditionally, full wave rectifier was considered better because of its lower output impedance comparing to the bridge rectifier. This was associated with that fact the only one diode is active at the moment. Also traditionally, bridge rectifier was considered better only because of the lower total cost of the implementation (demands only one secondary winding). During the time, full wave rectifiers almost disappeared.

Shortcoming of the full rectifier: in one (tube) application, circuit that was, using bridge rectifier, quiet regarding the hum, stopped to be so quiet with full wave rectifier, telling me something about the ground loops... Other than that, dynamics was improved.

Pedja

 

[moamps]

It's the secondary windings which takes abuse (more losses). I gather also that the iron core is not used at optimum. 2-diode solution with centertap is a budget solution but in most cases I think the extra losses are of minor importance.

Hi,

only one reason is using bridge instead of 2diode-double secondary in single PS; smaller overall price.
Additional secondary costs more than 2 diodes.

Regards

 

[fedde]

This is getting a good discussion!!!

Thanks for all the reactions. I'll keep you updated with my findings...

Fedde

 

[lohk]

Somebody at AA posted that he had opened an orginal Gaincard PSU and found an high quality R-core transformer inside and four fast 5A diodes. I do remember that everybody assumed that they are connected to form a normal bridge. What about if not? Did anybody take a closer look and coundt the secondaries?
Although I still do not like the idea of sharing rectifiers between channels (sorry Peter Daniel) would not make two full wave rectifier make more sense here ?

I will also try out the different possibilities soon (when time permits...)

Klaus

 

[Guido Tent]

This is getting a good discussion!!!

Thanks for all the reactions. I'll keep you updated with my findings...

Fedde

Hi Fedde

Did you consider the good old choke input filtering (regardless of the # of diodes 🙂

regards

 

[fedde]

Not yet 🙂

I do not have sources for good chokes too...

Fedde

 

[Guido Tent]

Not yet 🙂

I do not have sources for good chokes too...

Fedde

Fedde

I have them made for my XO supply boards

If you like I can ship one to try it (no charge, I think you are still studying ?), 400mH/100mAdc/4ohm.

Works fine with a 1000uF cap and about 50mADC load

regards

 

[keyne]

I second Guido's recommendation of choke input. It is the only way to not torture the diode. charge current peaks can be enormous and this causes noise. With the inductor you avoid the charge current peak, and the inductor takes out other noise that is there (like line noise, which there is a lot of)...
put an RC snubber on the output of the rectifier though to avoid high voltage peak when the diodes switch off (U= L*di/dt!).
In my opinion this supply topology, followed by an active regulator, should also be used for power amplifiers.
regards,
k