[rlee8394]

This provides partial rectification of the AC power before the rectifier tube and we have been told this will provide better service life for the tube.

    Actually the 1N4007 diodes are providing all of the rectification. Consider the operation in this quite elegant modification. When the AC single swings positive, one 1N4007, call it D1, is forward biased and thus conducts. The other 1N4007, D2, is reversed biased and does not conduct. When the AC signal switches, the diodes switch roles, D2 conducts while D1 is reversed biased.
 
    Now let's look at the 5AR4. Since it is in series with the two 1N4007 diodes, whenever one diode conducts, current will also flow through one of the 5AR4's diodes. The 5AR4 is a dual diode tube. What's interesting though is that while one 1N4007 and one half of the 5AR4 is passing current, the other half of the 5AR4 is idle. It doesn't have to handle the stress of the reverse voltage (PIV) as the 1N4007 is handling that function. Since it never sees the reverse voltage, one can assume that is can extend the life of the 5AR4. So then, as the AC cycle continues, only one diode within the 5AR4 tube conducts for each half of the AC cycle.

    This got me thinking. The forward resistance of a 5AR4 is about 100 ohms, thus producing a corresponding voltage drop across the tube rectifier. Not being one to allow perfectly good resources go to waste, I thought why just let half of the 5AR4 work at a time. If one were to parallel both sections of the 5AR4 together, then the internal resistance would be cut in half, thus resulting in a much less voltage drop across the rectifier. Since we know eliminating the reverse voltage from the 5AR4 is good practice, allowing both diodes to share the current duties, each diode in the 5AR4 only sees half the overall current passed through it. This would further extend the life and operation of the 5AR4 as well.

    Paralleling both diodes in the 5AR4 can be accomplished by connecting pins 4 and 6 on the socket for the 5AR4. This connects the two plates together, the cathodes are already connected. So as I see it, the benefits are:

1) The 5AR4 never sees reverse voltage.
2) Both diodes conduct simultaneously, improving current handling.
3) Less voltage drop across the diode for increased B+.
4) Still maintain the slow start capability for the amplifier.

    I am currently listening to one of my Ultimate 70's amps with this additional mod in place. I tried it with a NOS 5AR4 tube as well as two Chinese Shuguang 5AR4's. What's interesting is a few moths back, I made the mistake of turning the amp on "rapid fire" style which stressed on of the Chinese tubes to the point that when I turned it back on after several minutes, the diode started to arc over. Presumably because I stressed the PIV function. I lost track of which of the two tubes was the problem. I tried each of them after this mod only to find out that they both still worked! I can only surmise since the tubes never see any reverse voltage that the tube was holding up.

As always, YMMV, and make sure that you caps can handle the slight increase in B+.

Thanks,
Ron

[avahifi]

Interesting thoughs Ron.  I will try your suggestion with one of my Ultravalve amps today.

I will be concerned that the B+ voltage might go too high. Obviously the amp will need to be re-biased for the higher working voltage.

What happens if one of the 1N4007 diodes should short in this configuration?  Does this then increase the risk of 5AR4 failure or worse?

Regards,

Frank Van Alstine

[Tom Alverson]

Good thinking Ron.  The 5AR4 draws about 10W of filament power, and in a ST-70 about 4 Watts of "plate" dissipation using your 100ohm number.  This 4 watts would be cut in half by paralleling the tube sections.  You could add a series resistor after the tube and before the first capacitor to keep the voltage from rising (compared to the stock circuit).  This would probably help the tube reliability even more (than just adding the silicon diodes and paralleling sections) especially in those "hot" switching events where you turn it off and turn it back on quickly while the tube is still warmed up.  The added series resistor will limit the current surge where the hot tube tries to charge up the big power supply capacitors.  This web page even mentions adding extra series resistance to keep the tubes safe:

http://www.freewebs.com/valvewizard/fullwave.html

[rlee8394]

Frank,

    If a 1N4007 goes open, the you have half-wave rectification. If instead it were to short, then it would probably take out the other diode and blow the fuse since the diode would be across the secondary AC voltage. Either way, no damage to the 5AR4. You make a good point though. Since 1N4007's are both small and cheap, I'd probably use two or three of them in series just to provide a margin of safety in case one of them shorts. just like you did with the custom power transformer that you once offered for the SUper 70i.

Tom,

    Thanks for the link. I'll give it a good look.

    For those who may be interested, here's what I've done to my two Ultimate 70's. When I built them, I used a new Triode Electronics power for them which put out a bit more B+, and hence about 40 Watts/Channel. Even with tube rectification, the B+ was close to the limits of the 450 vdc caps. So I replaced the 68uF/450v cap with a 68uF/500 vdc unit and replaced the the second 68uF/450 vdc with a 560uF/450 vdc unit. I also wanted to drop the B+ a bit so I added an extra choke in series with the current choke. So after the rectifiers, there is 68uF > 1.5H > 1.5H > 560uF.  I could have added another cap between the two chokes, but that would keep the B+ high. So basically it's like using a single 3.0H choke which performed two nice functions; one, it lowered the B+ back to normal levels; and it also lowered the AC ripple by an order of magnitude. I feel that this extra choke really helped the sound. Could be my imagination, so I'm always looking for another set of ears to confirm or dispel what I perceive. Hence the disclosure of my mods. Note that these were done on the original kit board of the Ultimate 70 that Frank provides. I'll let Frank weigh in on the merits of performing these or any other modifications on its successor, the UltraValve.

   Speaking of perceptions, I'm currently listening to one of my Ultimate 70's with Franks dual 1N4007 diodes ahead of the paralleled 5AR4, the cap and choke mods mentioned above, and just for giggles and grins, I installed a set of EH KT-88's that I had from a Triode Elec. MK III kit. Now the KT-88's are uprated compared to the 6CA7/EL34 while maintaining the same plate-plate impedance (4300 Ohms) with increased power capability. The KT-88's are biased at 2.0 volts which equates to 65ma per tube. Turns out that if you turn the bias pots full counter-clockwise, that's where you end up roughly. I think I notice much more musicality, but like I said, I'd like a second opinion. I also have some 6BG6GAs which are very cool and are real cheap at $8 each from Mike at SND Tube Sales. You need adapters or have to rewire the sockets since they have a top anode cap. I also have a bunch of Russian 5881s with the thin coin base which I recently read that someone swears they have the best attributes of the KT-66 and EL34 without any of the drawbacks of each. Since this is a hobby, I'll try them both in my newly tweaked Ultimate 70's. It's nice having two of them for side by side comparison.

Sorry for the long tome, but I though someone may find this info interesting.

Thanks,
Ron

[Listens2tubes]

Great stuff Ron, keep it coming.

[simon wagstaff]

I am looking to do this upgrade over the holidays. I have the diodes in hand. So, is there a better way to do this or should I stick with the mod that is described originally?  I may need some photos if it is otherwise.

[Wayner]

There is a before and after wiring diagram on the 1st page of this thread.

Wayner

[simon wagstaff]

yep, I was referring to the mod just posted above or is it the same?  Anyway, I will do the one with the pictures....
 

thanks!

[Tom Alverson]

Try the original mod first (diodes in series with each power transformer lead to the rectifier tube).  Measure the voltage on the first power supply capacitor that connects directly to the rectifier.  If you want to try Ron's idea, short together pins 4 and 6 on the recitifer tube (ONLY do this if you have the diodes in series with the red transformer leads going to the tube).  Then remeasure the voltage to make sure it is still below the capacitor specs.

Tom

[rlee8394]

On the second unit that I modified last week, there was, as Frank had worried, an occurrence  of a shorted diode. Only the 3 amp SLO-BLO fuse blew. So no catastrophic failure. Therefore, I put two 1N4007 diodes in series to handle twice the voltage, 2000 volts instead of 1000 volts. So my advice is to use two or three diodes in series for extra protection. The diodes are pretty cheap, so don't skimp. You can use 1N4007, 1N5408, or 1N5399. The difference is current handling, not that you really need the extra current capacity. 1N4007 is rated 1.0 amp, 1N5399 rated 1.5 amps, and 1N5408 is rated at 3.0 amps. Your local Radio Shack should have these available.

Ron

[simon wagstaff]

Finally got the time to work on my U70 yesterday. It all went very smoothly following the diagram in this thread. Is there a difference in sound? I don't know for sure. It seemed like there was just a little more bass response, I had to turn the sub level control on my Infinity IM 4.1 down by 2/10 of a turn. Probably just me imagining things. Replaced the power cord as well. Had no problems figuring out which end to cut off.



Still sounds wonderful.  Had a great time  listening to music and watching the Jets win.

Today, more listening and hopefully watching the Ravens beat K.C.

I have an extra pair of diodes that I got just in case. If anybody wants them feel free to P.M. me.

[rlee8394]

Great stuff Ron, keep it coming.

   OK, I started to think about this a bit more. Since the 1N4007's are performing the full-wave rectification, the 5AR4, in addition to providing soft-start capability, just acts as a directional 100 ohm resistor. It ends up passing the pulsating D.C. to the first capacitor filter. So, I thought, why send pulsating D.C. through the 5AR4? How about filtering it first after the solid-state diodes. Then the 5AR4 just acts to drop smooth D.C. to the next filter stage.

   I cobbled together two 1000uF/250 vdc electrolytic caps in series, with bleeder/equalizing resistors, to create a 500 uF/500vdc filter cap and connected it between the 1N4007s and the 5AR4 to ground. This provides heavy filtering for the rectified B+ and, presumably, makes life a bit easier for the 5AR4 as it just drops the B+ a bit and doesn't have to deal with that pulsating D.C. I've been listening to it for about an hour or so. Sounds good. I think I hear better bass performance, and I suspect the overall sound quality has improved due to the addition reduction in ripple voltage. Of course, this is just subjective observation on my part based on theoretical engineering principles.

   I don't have an oscilloscope to make actual measurements, so I'll rely on Frank's group or you experienced techs out there to validate my observations. I don't recommend doing this unless the cap that you add, as well as the cap after the 5AR4 has a voltage rating of at least 500 vdc. With my Triode Electronics power trannie, I get 485 vdc across that 500uF/500vdc cap. That equates to nearly 60 joules of energy storage!! Compare that to 6 to 9 joules of storage for the 68uF or 100uF caps. You probably don't need 500uF for the filter between the 1N4007s and 5AR4. 100uF or better would work fine.

   So I can state that:

1) This seems to be a reasonable upgrade.
2) No harm has been done as long as you use caps of sufficient voltage.
3) With 500uF of capacitance, at turn-on, the power fuse doesn't blow.

   As always, mod at your own risk. If you have any doubts, then don't perform this or any other mods not approved by Frank on your Ultravalve/Ultimate 70 amps. Just enjoy them as they are!

Thanks,
Ron

[simon wagstaff]

Thanks for the tip on the speaker binding posts. Ordered some nice gold plated ones from the E-bay source you mentioned. 16 bucks for 4 sets of two (I needed a total of 4 red ones). Installation was easy enough though took a little time since I have sort of a small soldering iron. The existing speaker binding posts were the cheapest, nastiest ones around, my U70 is a kit built job. Fit with the bananas on my Audioquest speaker cables nice and tight now. Of course, no difference in sound but I think I will replace the RCA input jacks since they are the cheapiest as well. I can get some decent gold plated ones from the same source for about 7 bucks including shipping. I would like the fit to be a little tighter with the Audioquest cables I have running into the amp.

Did I mention I LOVE my U70?  I feel this is the last amp I will every need or buy.

[Brett Buck]

   I cobbled together two 1000uF/250 vdc electrolytic caps in series, with bleeder/equalizing resistors, to create a 500 uF/500vdc filter cap and connected it between the 1N4007s and the 5AR4 to ground. This provides heavy filtering for the rectified B+ and, presumably, makes life a bit easier for the 5AR4 as it just drops the B+ a bit and doesn't have to deal with that pulsating D.C. I've been listening to it for about an hour or so. Sounds good. I think I hear better bass performance, and I suspect the overall sound quality has improved due to the addition reduction in ripple voltage. Of course, this is just subjective observation on my part based on theoretical engineering principles.

   I don't have an oscilloscope to make actual measurements, so I'll rely on Frank's group or you experienced techs out there to validate my observations. I don't recommend doing this unless the cap that you add, as well as the cap after the 5AR4 has a voltage rating of at least 500 vdc. With my Triode Electronics power trannie, I get 485 vdc across that 500uF/500vdc cap. That equates to nearly 60 joules of energy storage!! Compare that to 6 to 9 joules of storage for the 68uF or 100uF caps. You probably don't need 500uF for the filter between the 1N4007s and 5AR4. 100uF or better would work fine.

    That's an interesting idea. Did you determine the inrush current? I would think that a discharged 500 mfd cap attached across the diodes looks like a dead short for a while - I would expect more than the 1 amp rating of the 1n4007. It's certainly on the wrong side of the "soft start" unless I am missing something.

   I would also be very nervous about 485 volts across a 500V rated cap, but that would be easy to fix (just use two 450V caps in series instead).

   Brett

[rlee8394]

Brett,

The transformer only has a 300ma rating, so the 1 Amp diodes are fine. The 485 volts on the caps are well within the 500 volt rating and well below the surge voltage rating. I've been running this daily for 6 + hours a day with no problems. Use 450 vdc caps for extra margin of safety if you wish. Like I said, I just cobbled these together because I had 1000uF/250 vdc caps. I'll probably go with 350-450 volt rated caps when I perform this for good. The caps charge up pretty fast, so whatever the inrush is, it doesn't last long.

I just did the same"hybrid rectification" mod on my Super PAS 3. I did voltage checks on it, but haven't listened to it yet. I had a couple of dual JJ 100uF + 100uF /500v caps, so I wired the two + terminals together for 200uF/500v after the SS diodes.

Ron

[Listens2tubes]

Ron, you could post this mod in the Tube-O-Phile forum for those who are less fortunate in not having Frank's products. Spread the love, fellow tweaker.

[Brett Buck]

The transformer only has a 300ma rating, so the 1 Amp diodes are fine.

  Continuous, yes. The only thing stopping it from going above that is the impedance of the secondary/the charging rate of the capacitor, near as I can tell.   I will try to figure out what it wants to be when I get a chance.

    Brett