what's your opinon on using HexFred diodes in tube amps?

Discussion in 'Amps/Cabs Tech Corner: Amplifier, Cab & Speakers' started by jvegas, May 4, 2008.

  1. jvegas

    jvegas Member

    Messages:
    3
    Joined:
    Apr 23, 2008
    I was thinking of using some HexFred diodes in the power supply section of my 5150. I heard this gives a positive effect on the sound of the amp ,better bass, less harsh mids and highs and all around better clarity. I like to have some more opinions on this matter.


    J.
     
  2. Structo

    Structo Member

    Messages:
    9,573
    Joined:
    Apr 10, 2007
    Location:
    Oregon
    The only advantage I have read about is less switching noise.
    Most can't hear the difference.
     
  3. Trout

    Trout Member

    Messages:
    7,576
    Joined:
    Jun 23, 2006
    Location:
    Illinois/Far West Burbia
    On the opposite end of the spectrum, I have often wondered if some of the newer Fast soft recovery rectifier diodes would better emulate a tube rectifier. Fast switching but soft response.

    I noticed these a bit back, but have not located a good source for them in smaller quantities.

    Rectifier Diode Data
     
  4. mooreamps

    mooreamps Senior Member

    Messages:
    377
    Joined:
    Nov 8, 2007
    Location:
    Sunnyvale, Ca.
    Most probably can not hear the difference as using the fast recovery diodes only removes one of the hiss makers in a pre-amp. The thermal noise generated by using carbon resistors in the cathodes of the pre-amp are also hiss makers. As such, I use fixed bias for pre-amp gain stages, instead of self bias, and the fast recovery diodes in my pre-amps to totally abate the hiss noise.

    -g
     
  5. Trout

    Trout Member

    Messages:
    7,576
    Joined:
    Jun 23, 2006
    Location:
    Illinois/Far West Burbia

    That is actually a pretty good idea, what percentage of hiss is removed on the average? Did you get any scope readings on this?

    Also, any experience with the fast soft recovery rects? I always suspected them to reduce perceived noise as well.

    Trout
     
  6. donnyjaguar

    donnyjaguar Member

    Messages:
    4,243
    Joined:
    Nov 4, 2005
    Location:
    Canada
    This comes up every once in a while. Using Hexfreds, or any "fast recovery" diode in a 50/60Hz application is a total waste of time and money.
     
  7. stratman_el84

    stratman_el84 Member

    Messages:
    403
    Joined:
    Mar 16, 2008
    Location:
    Kalamazoo MI area.
    I gotta agree with you Donny. I simply can't see the advantage of fast/soft switching diodes in a 50/60Hz rectifier circuit. Normal diodes' switching times are already far far faster than the relatively glacial pace of switching 50/60Hz AC power for rectification in a power supply.

    The only really justified use I can see for those diode types in a guitar amp would be for low-level signal path switching.

    Cheers!

    Strat
     
  8. jay42

    jay42 Member

    Messages:
    6,800
    Joined:
    Dec 20, 2006
    Location:
    Sandy Eggo
    You're trying to avoid inter-modulation. Yes, the frequencies of the noise coming of PS diodes are above the guitar amp range, but it can intermodulate.

    So Mr. Moore, there are diodes in your pre-amps? Are you talking about generating distortion, keeping the power amp from sagging the pre-amp power supply, or diodes in your fixed bias scheme?
     
  9. Fretts

    Fretts Member

    Messages:
    363
    Joined:
    Nov 20, 2003
    Location:
    Santa Monica, CA
    My understanding is that it isn't the Fast Switching that is the benefit, but the soft recovery. When the diodes are changing state from conduction to off, and vice versa, they can produce a spike that in turn can produce other effects in the power supply that can be audible. Soft recovery diodes, are capable of changing state with a more 'rounded corner' if you will, and are thought by some to reduce eliminate one more source of noise.
     
  10. Hwoltage

    Hwoltage Member

    Messages:
    9,560
    Joined:
    May 6, 2010
    Location:
    Corvallis, Oregon
    You can lessen the switching noise by placing a 10 ohm or so resistor just after the DC feed of the rectifier but before the reservoir cap. The wattage is the total current of the amp squared and multiplied by the ohm rating of the resistor. Much cheaper and also protects your caps a bit as they aren't slammed so hard by the B+ upon startup of the amp.
     
  11. Chrome Dinette

    Chrome Dinette Member

    Messages:
    14,435
    Joined:
    Aug 4, 2005
    Location:
    Pittsburgh, PA
    I have done this and didn't really notice much of a difference, if any. Since they are cheap, I end up using UF4007's instead.
     
  12. teemuk

    teemuk Member

    Messages:
    2,876
    Joined:
    Apr 21, 2009
    And of course there's the old trick of bypassing the diodes with ~ 100 nF capacitors.
     
  13. Hwoltage

    Hwoltage Member

    Messages:
    9,560
    Joined:
    May 6, 2010
    Location:
    Corvallis, Oregon
    So that's what those are for!! Is this for a delay? Would you want to bypass the caps then as well? At least the reservoir? Wait - do you mean pf?
     
  14. teemuk

    teemuk Member

    Messages:
    2,876
    Joined:
    Apr 21, 2009
    It slows the diode's operation, yes, but at the same time it will help to damp the flyback bursts caused by diode's switching interacting with leakage inductance of the PT. As a secondary effect it lowers the ringing frequency therefore decreasing the amount of EMI and that "diode switching hash" you will hear as noise that creeps to the signal. Yes, the value is nanofarads. Typically anything between 10 nF - 100 nF.

    So, in ordinary linear supplies both super fast diodes, or intentionally super slow diodes (damped by bypass caps) can achieve the same thing. SMPS are a different matter. They would self destruct with slow diodes.

    Oh, a third alternative is to use a toroidal PT, which has a lower leakage inductance than typical E-core transformers, but that's way more expensive than four 10 - 100 nF caps.

    A good ground return noding is also essential. You don't want the power supply filter ground to interfere with signal grounds because the filter ground will conduct huge current bursts caused by capacitors charging from the energy the rectifier pumps into them.

    I don't see reasons for bypassing filter caps unless they have a poor ESR or unless you for some reasons need to damp the tank circuit of the power supply (big filter caps may have considerable series inductance). In a good design there should be no need for that, SMPS and power supplies of class-G amps may be a different matter. Oh, those 100 nF caps you often see scattered here and there on the power supply rails typically are not either but compensation against wiring/trace inductance.
     
  15. VacuumVoodoo

    VacuumVoodoo Member

    Messages:
    1,567
    Joined:
    Jan 9, 2005
    Location:
    Gothenburg, Sweden
    This is why you don't want only a capacitor parallel with the diode, you want to significantly decrease Q of the resonant circuit formed by the capacitor and PT leakage inductance. What you need is a ca 10k resistor in series with each of those capacitors. Makes the difference between passing and failing EMC test on radiated RF interference.
     
  16. Blue Strat

    Blue Strat Member

    Messages:
    30,036
    Joined:
    Aug 2, 2002
    Location:
    Sterling, VA (not far from Washington DC)

    Sounds compelling, but so does the adage "(different tube) rectifiers (of the same type) can't make a difference...they're not in the signal path...they're only in the power supply".

    Point being, often, what seems logical and "proven by science" is often incorrect due to our incomplete understanding of the situation or missing one of the several important details. It happens in all fields, by people of all skill/education levels.

    This is why empirical evidence rules all. Since it's not difficult or expensive to obtain empirical evidence, it's the preferred approach AND why it's included in the "scientific method".
     
  17. teemuk

    teemuk Member

    Messages:
    2,876
    Joined:
    Apr 21, 2009
    ^ That's sort of like saying that if you get kicked in the groins with a motorcycle boot it might not hurt you because there could be several important details in the human pain sensing mechanism we are missing or fail to measure. In practice, we pretty much know it will hurt like hell even without testing.

    The thing with rectifier diodes in a low frequency power supply is about the same. Some things are just common sense. If you don't believe me, take Nelson Pass' word for it:
    (http://www.hometheaterhifi.com/volume_4_2/nelpass.html)

    On that note, the idea of using faster recovery rectifier diodes in the power supply (whether they are Schottky HexFred or whatever) has beed discussed to death in DIY hifi and high end audio forums because those diodes are one of those itsy bitsy additions many people swear by - as if the rectifer diodes would make a difference between a mediocre and great amp. Those who bothered to test them (instead of just being convinced they have some sort of effect) have measured no significant improvement in audio quality or noise level. If noise / hiss is an issue you have several ways to compensate it (as we already learned from posts in this thread). All that without even touching the rectifier diodes. Chance is, if there is an issue, the issue is way bigger than the diodes.

    They make total sense in SMPS applications, much less in low frequency linear power supplies. But anyway, buying a handful won't drive anyone to bankruptcy so if using them soothes your mind then equally as you don't really gain anything significant from using them you won't be losing anything either.
     
  18. Blue Strat

    Blue Strat Member

    Messages:
    30,036
    Joined:
    Aug 2, 2002
    Location:
    Sterling, VA (not far from Washington DC)
    That's exactly the opposite of what I said. Your example is the EMPIRICAL proof that I was talking about.

    And you're proving my point again. "Common sense" to who? Someone with no electronic background? To someone who learned everything he knows on internet forums? To a BSEE? Who? Common sense is meaningless in a scientific, and many other, situation. There's theory, and there's empirical proof. All else is BS.
     
  19. Hwoltage

    Hwoltage Member

    Messages:
    9,560
    Joined:
    May 6, 2010
    Location:
    Corvallis, Oregon
    I have been building my grounding layout first with the rectifier neg and res cap neg to the same node, that node to the chassis. From that node to another node where I group the speaker jack ground (from an isolated jack), the grid leak and cathodes of the output tubes, and the neg of the second filter cap. From the neg of the first cap I run either a line of wire that has more resistance than the wire used for the ground of the output section, or 5 or 10 ohm resistor to a third node where the neg of the 3rd cap sits. I group all the grid leaks and cathodes of the PI together and run them to that node and so on down the line. The only physical connection to the chassis is back at the rectifier ground and all the power taps needed for each stage (usually one stage per filter cap) are connected right at the lead of the cap. The idea is that it's then impossible for any ground current of later stage to find it's way home higher up through the preamp, which creates a larger loop and more radiated interference. Taking a lead from the neg of the first cap with more resistance to the preamp than to the chassis pretty much assures that the current loop for the PI doesn't stray into the grids and vice versa. At least that's where I'm at now.

    As far as the cap/resistor for the diodes - to find the impedance that the transformer would see due to the cap and resistors -

    Does each side of the transformer, as it feeds to the rectifier on each side, see two paths to ground consisting of the impedance through the cap and resistor straight to ground at the rear of the rectifier at shutoff and also through the impedance created by the resistor and cap with the ESR of the first cap added in series, then to ground, at forward bias? If this is the case once I have the vector quantity of the impedance do I add the two impedance values off the back of the rectifier in parallel and the two impedance values through the cap in parallel and assign one to each side of the transformer accounting for both cycles or - do I add one impedance off the back of the rectifier to ground in parallel with the impedance to the cap and to ground then assign a value to each side of the transformer? This makes more sense considering how a rectifier works... Am I even close?
     
  20. pdf64

    pdf64 Member

    Messages:
    5,178
    Joined:
    Aug 9, 2008
    Location:
    Staffordshire, UK.
    'What you need is a ca 10k resistor in series with each of those capacitors'

    10k? 47 - 470 ohms ia a more usual values for snubber circuits, assuming that I've got the right end of the stick.
     

Share This Page