Tube DIY Asylum

Do It Yourself (DIY) paradise for tube and SET project builders.

Return to Tube DIY Asylum

Message Sort: Post Order or Asylum Reverse Threaded

Mercury Vapor voltage drop.

67.189.207.135
Posted on March 30, 2017 at 16:01:40
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001
 a bunch of posts below gusser commented how MV tubes had less "sag" that other rectifiers and I responded they don't have any sag but a constant voltage drop. Having never seen Vf vs. current plots I wasted a few minutes creating a crude set.





So the voltage drop doesn't appear to be a precisely fixed voltage but i have to say it seems to be pretty constant. This leads me to question how to "qualify" the voltage drop across a MV tube. Since it is close to constant over a pretty wide range of current I want to say it has low to almost no impedance and the voltage drop is just that... a fixed drop, does this mean it just gets put in the loss column and forgot about?

Up next tungars :-)

dave

 

Hide full thread outline!
    ...
Why use Mercury?, posted on March 30, 2017 at 16:41:33
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001
 aside from the really yummy sound....500V @ a quarter of an amp...







.






.






.




 

Will be studiously studying your 'tungars' results-nT, posted on March 30, 2017 at 16:46:17
Cleantimestream
Audiophile

Posts: 7550
Location: Kentucky
Joined: June 30, 2005
 !
The Mind has No Firewall~ U.S. Army War College.

 

Consider it a "reverse biased" voltage source?..., posted on March 30, 2017 at 16:57:40
Steve O
Audiophile

Posts: 12362
Location: SE MI
Joined: September 6, 2001
 ....at least over the constant drop region.

Edit: In series with a diode, of course.

 

RE: Why use Mercury?, posted on March 30, 2017 at 19:13:54
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 Series wired pairs of 1200 PIV Schottky diodes, with voltage equalizing resistors, will get the job done. No HAZMAT and no "hash" with the Schottkys. :>D


Eli D.

 

RE: Why use Mercury?, posted on March 30, 2017 at 20:27:27
Cleantimestream
Audiophile

Posts: 7550
Location: Kentucky
Joined: June 30, 2005
 Hash is never heard at our listening frequencies, Eli.

The superior sound is why I use them, however if I am using mainstream efficient speakers 85-90 DB speakers, one cannot hear the superior detail and subtle nuance in the bass region.

At least I am unable to through my Proac tabletts nor the Infinity reference 4's I had.
The Mind has No Firewall~ U.S. Army War College.

 

RE: Why use Mercury?, posted on March 30, 2017 at 21:12:43
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 I suspect the absence of "sag" compared with vacuum rectifiers is what you like when Hg vapor is employed.

No sag with Schottkys too and they are "dead" quiet. The time delays you use with Hg vapor will work without change, in combination with high PIV Schottkys. Give 'em a try, if you have not already done so.


Eli D.

 

RE: Why use Mercury?, posted on March 31, 2017 at 07:44:37
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001
 about 20 years ago I made up a "mercury vapor fredifier" which consisted of a dead 71A busted open and a pair of hexfreds soldered in place and a couple of blue LED's. A bit of matte spray on the inside of the envelope opbsured the innards enough to make a pretty convincing 83 replacement. At the time I was using a Loesch type linestage and could sub between different rectifiers very quickly. The swap du jour of that timefram was between three different "mercury" type tubes, 83's 287a's and the fredifier. Nobody had a clue that one of them was a sand device and at the time the hexfred was to be the sure death of any future use of tube diodes. Long story short, the difference in sound was easily heard with several people mentioning the "hash" they heard from the fredifier.

I guess it has been 20 years and I currently have a phono stage that can take dampers or mercury and I use those particular schottlys in my quad HV supplies so I guess I'll report back.

dave

 

RE: Will be studiously studying your 'tungars' results-nT, posted on March 31, 2017 at 08:41:59
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001




OK... Tungar vs. Tunmerar.

for the sake of interest, Tungar is short for Tungsten Argon and the particular mercury tubes also have argon in them which must make the mercury inert in the case of breakage :-)





 

RE: Why use Mercury?, posted on March 31, 2017 at 09:41:19
PakProtector
Audiophile

Posts: 12363
Joined: May 14, 2002
 Those 4E27 amps you saw at The Mill ran || 6AU4's into an LC supply for 590V at 255 mA. The 6AU4's will be replaced with 866AX when next I have the hoods up...:)

Douglas

Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.

 

RE: Why use Mercury?, posted on March 31, 2017 at 09:46:49
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 FREDs make a small amount of switching noise, but that which they do produce can be highly obnoxious. My preferences in SS diodes are Schottkys and UFnnnn (in that order).

Jim McShane has a "love affair" with the GI856. Sadly, that part has been discontinued. The GI856 does sound good. Both the amount and nature of switching noise seem to matter. Obviously, no noise Schottkys rate high.


Eli D.

 

RE: Why use Mercury?, posted on March 31, 2017 at 10:51:23
tube wrangler
Manufacturer

Posts: 2484
Location: USA
Joined: January 29, 2007
 The blue Leds and the matte spray.......

Their sonic effects were.......

-Dennis-

 

RE: Will be studiously studying your 'tungars' results-nT, posted on March 31, 2017 at 12:55:25
sideliner
Audiophile

Posts: 208
Location: NYC
Joined: August 22, 2013
Contributor
  Since: December 15, 2023
 Dave,

Thanks for the graphs. Do you have a subjective preference between these two? Also, have you tried any xenon rectifiers such as the 3B28? I'm only asking because if these noble gas tubes sound good, they'd be a nice worry-free alternative to the MV rectifiers.

 

Schottky, posted on March 31, 2017 at 13:03:11
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 Vishay refers to the GI856 only as a fast switching rectifier. Do you have a "go-to" HV Schottky besides the Cree? That tab is a PITA.


 

RE: Will be studiously studying your 'tungars' results-nT, posted on March 31, 2017 at 13:39:14
dave slagle
Manufacturer

Posts: 5430
Location: NYC
Joined: April 27, 2001
 I haven't done a direct comparison of the tungar to MV becuase their applications are quite different. The MV tubes typically are high voltage / low current (500V-1000V @a few hundred ma) and the tungars are low voltage high current (300v @ 8A). This puts the mercury in my amps and the tungars in my field coil supplies.

I have tried the 3B28 and the are a drop in replacement to 866's like ripple is a drop in replacement to wine.

dave

 

RE: Schottky, posted on March 31, 2017 at 14:27:58
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 Except for surface mount variants, all high PIV Schottky diodes are TO220 or TO247 cased. :>(

What you can do is buy higher forward current rated parts and saw the tabs off.

I too wish the high PIV Schottkys came with axial leads, but SMPS and the like are the targeted market. We get to go along for the ride.


Eli D.

 

RE: Schottky, posted on March 31, 2017 at 14:45:11
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 I think we might have discussed this before. Sorry, it just came back to me. The diodes Jim likes are still available from at least one supplier on Amazon, and they're not too expensive. They have a Trr of 200nS, much faster than the typical 30uS seen with diodes like the 1N4007. Suppression caps are probably still required though.

I guess we should count ourselves lucky that the Crees are still available in through-hole. Someday, they'll probably all be SMT only.




 

RE: Schottky, posted on March 31, 2017 at 18:20:44
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 IIRC, the UF5408 is somewhat faster than the GI856. Mere speed is not the only puzzle piece. AFAIK, FREDs are the fastest PN junction diodes, but not necessarily the best sounding. The character of the reverse recovery spike and the ease of suppression also matter.

Jim uses series wired GI856 pairs and a single film snubber cap. in the PSU of a H/K Cit. 2. The result is excellent. Of course, the low DCR choke after the doubler stack definitely contributes. A "hash" filter, made from "RF" parts, located between the doubler stack and the main filter choke yields an additional slight improvement in background "blackness". The "hash" filter kills crud from 3 potential sources: AC mains "hitchhikers", SS diode switching noise, and ripple overtone energy associated with the large valued doubler stack capacitors.


Eli D.

 

RE: Schottky, posted on April 1, 2017 at 10:46:27
Jim McShane
Dealer

Posts: 5910
Location: Chicago
Joined: January 13, 2003
 My affection for that part is based on the excellent diode testing article done for Audio Amateur a number of years back. Rick Miller (who has excellent credentials but I can't recall them off hand) showed clearly the near total lack of PN burst noise in the GI diodes even when compared to "fast" diodes. He pointed out that speed was not the only criteria for a high performance diode - "soft" turn off was another critical element.

He also pointed out that the diode performed just as well or better without snubbers.

Finally both he (IIRC) and John "Buddha" Camille (for sure, it was published in Sound Practices) pointed out how certain diode turn off characteristics could excite noise in the secondary of the power trafo. John recommended a couple snubber designs, but NOT the common cap bypass of the diode iteself. I highly recommend his article - it's probably on line somewhere.

 

RE: Schottky, posted on April 1, 2017 at 10:49:35
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 Would that filter be part of Jim's "Level II Plus" upgrade? I'd be interested to see the component values, but maybe Jim considers this proprietary info. In any event, I've been thinking along the same lines for my own HK doubler rebuilds (A500 & A700 types). Installing larger value doubler caps and higher current rectifiers is a given, but I'm on the fence about the post-filter treatment. I have a stash of small 500mH / 30 ohm chokes, but I'm not sure yet whether they're optimal for this application (I bought them for use in non-doubler supplies). Anyway, I suspect the limited space in the Cit II would dictate something around 10mH or less, which would also better fit your description of "RF" components. Do the series diodes have equalizing resistors in Jim's design?

I've also been wondering whether it might be effective to configure the doubler filter as a CLC. In other words, the doubler caps would be followed by a small choke, than another filter bank. The second bank would be smaller, just enough to provide isolation from the DCR of the choke for short-term peaks and to ensure that the supply performs as a voltage source at higher frequencies. Luckily, all my rebuilds will be on a new chassis, so space constraints will be only what I make them. :)


 

RE: Schottky, posted on April 1, 2017 at 11:43:50
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 Thanks for that info Jim. I located a General Semiconductor PDF on this topic that refers to Rick's article (Audio Amateur - Jan '94), but not the article itself. I get the gist of this though. Your experience speaks for itself; I'll take this approach for future builds.




 

CLC / Snubbers, posted on April 1, 2017 at 12:11:07
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
I just want to add something to my post above regarding the CLC concept. Even when it consists of only a small choke after the main filter, then a snubber, the effect on the PS can be severe. Below is a SPICE simulation of such a design. Note how the output Z of the supply spikes up at about 3.3kHz. The difficulty in dealing with this is one of the reasons I'm undecided regarding this approach.




 

RE: Schottky, posted on April 1, 2017 at 12:44:46
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 I think it's fair to say that the "hash" filter idea is in the public domain. L_RD knows that I've suggested it on numerous occasions.

The filter is simply a LC section made from what would ordinarily be regarded as RF parts: a 1000 pF. cap. either in mica or NP0/C0G ceramic combined with a high current RF choke. Use the choke with the largest inductance that does not saturate under the PSU's load demand.

Mouser carries suitable Bourns/J.W. Miller and Vishay parts. Look into the 5900 series parts on the linked catalog page.

No diode voltage equalizing resistors are in a "McShaned Deuce".


Eli D.

 

RE: CLC / Snubbers, posted on April 1, 2017 at 13:03:21
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 Run a simulation with CI/P - LChash filter - LCreservoir topology. Use plenty of capacitance in the I/P and reservoir positions.

The "hash" filter kills crud that might sneak into the reservoir cap., via the principal choke's winding capacitance. The big caps. crush the ripple fundamental. The other stuff deals with the inevitable ripple overtones and other noise.


Eli D.

 

RE: CLC / Snubbers, posted on April 1, 2017 at 13:19:21
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 "Run a simulation with CI/P - LChash filter - LCreservoir topology."

Well, yes, if a large amount of reservoir capacitance is added at the output, it attenuates almost everything. Is that the technique we're discussing? I thought you were talking about adding a RF hash filter to the existing doubler in the Cit II. Do Jim's mods include a hash filter plus another large bank of capacitance? That certainly makes the task easier. I just assumed there wouldn't be space for all that.

 

RE: CLC / Snubbers, posted on April 1, 2017 at 14:23:35
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 It all fits. :>D He uses 820 μF. caps in the doubler stack. The principal filter choke is the excellent OEM part. Jim gains space on the cap. bracket by moving the bias supply off of it. Capacitance galore in the doubler, reservoir, and decoupling positions. Something in excess of 130 WPC pulse is available. That PSU has brass monkey cojones. The "hash" filter is a small refinement of his already very nice work.

The linked cap. bracket photo is somewhat dated, in that parts brands have changed over time. The pair of isolated gray caps. are the doubler stack.


Eli D.

 

RE: CLC / Snubbers, posted on April 1, 2017 at 14:48:21
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 According to my sim, the original Cit II PS has a significant impedance bump around 25-30 Hz. That's due to the factory values for the choke and reservoir cap. Making the reservoir larger mitigates that issue. Anyway, this is very different than what I thought you were describing when you mentioned adding a hash filter. The model I posted is valid for that, but it doesn't include a large smoothing capacitor (reservoir) afterwards. I might have to bite the bullet and take the same approach in my rebuilds. The 500mH chokes will work if I do that, it's just a matter of adding enough output capacitance to push the resonance bump to a lower frequency. Good discussion, thanks!

 

3B28s are a drop in replacement to 866s like ripple is a drop in replacement to wine, posted on April 2, 2017 at 06:28:17
Jim Dowdy
Manufacturer

Posts: 1518
Location: Atlanta, GA
Joined: July 22, 2000
 What a great analogy!

Thanks for the graphs - I've been curious about it for years.

Jim

 

Hash Filter / Diodes , posted on April 2, 2017 at 07:47:32
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 I modeled this last night, and the PRV in each leg never exceeds the rating of a single diode. It's approximately the same as the output voltage (2.8 X RMS). I guess that's why equalizing resistors aren't strictly necessary when two are used in the Cit II. The A500 used single 500V diodes to produce +370V, and the A700 used single 600V diodes for an output of +410V. Today's higher line voltages squeeze those windows somewhat, so I will probably use two diodes in each leg of the A700 rebuild. The A500 can use 600V singles.

The general values you're describing for the hash filter - e.g. 0.001uF and 1mH - produce a huge output Z spike between 100kHz and 200kHz. The supply effectively turns into a current source at that frequency. I don't know whether this could have an impact on the power amp or other stages, but a 10uF electrolytic across the 0.001uF will flatten it.

Good discussion, my rebuild path is becoming clearer!


 

Hash Filter Model, posted on April 2, 2017 at 08:03:50
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 Just wanted to add this image so you can see that the ultrasonic spike isn't only a small anomaly. If the goal of the hash filter is RF suppression, this doesn't do the job! Granted, it's well out of the audible spectrum, but any energy on the B+ line at that frequency could cause the PS voltage to swing like crazy. 10uF across the .001uF resolves this completely.




 

RE: Hash Filter Model, posted on April 2, 2017 at 09:57:07
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 Go for it! My objective was to block noisy crud, regardless of source, from sneaking into the reservoir capacitor. There is always room for variations on a theme.

FWIW, I'd still like to see a simulation with a LC reservoir section located after the "hash" filter. Superior designs, like the "Deuce", some McIntosh models, and some Marantz models, include the LC reservoir section following the doubler stack. The choke is not particularly large valued. When big caps., ala McShane, are used at the I/P of a PSU filter, the ripple waveform is highly "triangular". Applying Fourier's Theorem to that "sharp" waveform led me to call for "hash" filtration immediately following the I/P capacitance. Energy well up into RF "territory" is present that can get past the principal choke, via its real world winding capacitance. Kill the garbage before it gets to the "back door".


Eli D.

 

Cit II Doubler Filter, posted on April 2, 2017 at 14:52:02
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 Here's the model of the factory CIT II filter. I had to guesstimate the DCR of the choke and ESR of the caps. I'll be glad to change any of the values or insert a hash filter using your components. Output Z of this design isn't particularly good, probably swings quite a bit when a 30Hz note is present. Then again, it was only $160 in kit form. :(






 

RE: Cit II Doubler Filter, posted on April 2, 2017 at 15:38:49
Eli Duttman
Audiophile

Posts: 10455
Location: Monroe Township, NJ
Joined: March 31, 2000
 I've snipped from Jim's site the minimum level of Cit. 2 PSU overhaul he offers. Jim offers stuff beyond this.

For the Citation II Power Supply

1. (2) 820 uf 250 volt 105 degree Celsius very low ESR clamp mount caps that increase the capacitance from
100 uf total in the doubler to 410 uf. (2) clamps are included.

2. (4) General Instruments fast/soft recovery diodes that replace the existing top hat diodes. They fit nicely in the
fuse holder right where the old top hats are now. The GI diodes are very good, and quite reasonably priced too.
If you increase the capacitance, you really need to replace those diodes. And the DO-41 packaging is ideal for replacement.

3. (1) JJ 100/100 dual section 500 volt clamp mount cap, and...

3a. (1) Nichicon 220 uf/500 volt clamp mount cap. The Nichicon replaces the 40 uf 525 volt original twistlock,
and the JJ replaces the dual 50/50 V1/V4 decoupling cap. (2) clamps are included.

Note: This amp responds very well to improved decoupling of the input 12BY7A tubes!

4. (1) .01uf high voltage film cap to replace the existing ceramic unit.

5. (1) Gen'l Instruments fast/soft recovery diode to replace the bias selenium rectifier

6. (1 ea.) 220uf and 100 uf radial electrolytics, a new 1.8K ohm resistor, and a terminal strip for the bias circuitry.
Two new 4.7K ohm resistors to increase the range of bias adjustment.

7. A copy of the schematic, and kit installation notes and diagrams.

You end up with 630 uf (vs. 140 uf stock) of high quality capacitance for the B+, and twice the decoupling
capacitance.


Eli D.

 

RE: Cit II Doubler Filter, posted on April 2, 2017 at 20:10:46
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 Yes, Jim's basic mods to the doubler create a much better supply. I modeled it yesterday, based on the info on his Website. It's not likely I'll ever own a Cit II, but the concepts Jim is using would be applicable to any supply of this type. Below is the model of the modified doubler. It's scaled the same as the factory PS I posted earlier. Quite a difference!






 

RE: 3B28s are a drop in replacement to 866s like ripple is a drop in replacement to wine, posted on April 4, 2017 at 12:39:55
sideliner
Audiophile

Posts: 208
Location: NYC
Joined: August 22, 2013
Contributor
  Since: December 15, 2023
 Thank you Dave for sharing your thoughts on the 3B28, never had a chance to hear it myself, but would've been nice if it was a sonic match to the MV rectifier. BTW, I had a good laugh with your ripple/wine analogy - good one.

 

RE: Cit II Doubler Filter, posted on April 4, 2017 at 13:02:13
Jim McShane
Dealer

Posts: 5910
Location: Chicago
Joined: January 13, 2003
 Hi TK,

Thanks for running the model! I know you didn't have all the best info to run it with either - so your effort was greatly appreciated.

Just FYI - the DCR of the Cit II choke is 11 Ohms at room temperature - it may be a bit higher when under the hood of a Deuce which does get warm underneath.

Also the ESR of the doubler caps is .187 Ohms each - so even stacked in the doubler the total ESR is under .4 Ohms.

Finally (again just FYI) there is 220 uf of high quality capacitance after the hash filter so the spike you saw isn't there.

One last time - THANKS!!

 

RE: Cit II Doubler Filter, posted on April 5, 2017 at 08:21:30
Triode_Kingdom
Audiophile

Posts: 10044
Location: Central Texas
Joined: September 24, 2006
 NO need to thank me, it was a necessary exercise for my own work. I was glad to see such a remarkable improvement from your upgrade design (not that I expected anything else). I have a number of HK amps and receivers to rebuild using the transformer sets on new chassis designs. Most will probably be sold off, but One of these, the A500, will be for my own use. I just picked up a spare power transformer for this amp, so now I'm thinking about re-configuring as two isolated channels or maybe monoblocks. Anyway, I'll probably be looking to you for the filter caps in some or all of these. The extremely low ESR of the components you're selling has really caught my attention. Send me a PM if you ever have any pending designs you'd like to see modeled in SPICE off the forum.

 

Page processed in 0.036 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]