[Bill Epstein]

Finally got my latest Cornet build finished this AM.

With 600volt, 4 amp Fairchilds and a 1.2 ohm R223 I got 4.9 vDC on the filaments. I replaced the resistor with a jumper and now have a whopping 5.25!  Both legs of the supply yield 3.6volts AC at R103 so it has to be the diodes. I've used Cree Schottkys lots of times and hadn't ever had this kind of voltage drop; live and learn.

What's kinda ironic is that I borrowed the specified 1N5821s for DC filament service in my recent octal Aikido build and I didn't need a series resistor at all.

On the bright side, if you need 5 volts for a 300B the Fairchilds could be used in place of a filament transformer

[jameshuls]

I have 4.3V on my supply, so I wonder what will happen when I install the IXYS HiPerFREDs (15A 1200V) in my Cornet... Perhaps it will balance out just fine?

[tubesforever]

Bill,  I used the 4amp 600v Fairchild Hexfreds from Percy Audio.  These were $1.65 each I think. 

The DC volts on my Cornet 2 were 6.29  so I used a 1/4 ohm 5 watt resistor to get to my favorite 6.1 volts DC.

On my Clarinet I am getting 6.0 volts with a straight jumper wire.  BTW I measure these after about 10 minutes bench time.  The voltages are low while the things heat up a bit.  They really do perform very well.  

I wonder exactly what is going on with the hexfreds vs the Schottke diodes.  The Schottke diodes delivered far more voltage.  I like the sound without the step down H+ resistor better with my Clarinet.  So this is simply a nice side benefit of the hexfreds--that plus a noise floor that defies description.  I simply cannot hear the floor because my amp is many times louder. 

Jim did good with his design work!

Happy Holidays Bill! 
BTW How is the critter sounding?

[hagtech]

Are the Fairchilds "schottky" type?  The circuit is designed for the low voltage drop of schottkys.  Normal diodes will cause too much drop.

jh

[tubesforever]

Jim,

From my understanding HEXFRED's are high speed epitaxial soft recovery diodes.  Here is a link to some information on them at Michael Percy's website.  There is also information on hexfreds on the Digikey and Mouser sites.

http://www.percyaudio.com/Catalog.pdf  (The Hexfreds are on page 9.)

Here is another from datasheetcatalog.com

http://www.datasheetcatalog.com/datasheets_pdf/H/F/A/0/HFA04TB60.shtml

I have a friend at Boeing who is big into DIY and is a power supply afficianto.  He dislikes zener diodes.  He has recommended I use the stealth Fairchild HEXFREDS for DC supplies because they are many times more quiet and are non ringing.  He says most of the 25k and up Hi End reference amps are using Hexfreds for DC conversion. 

Your equipment with the Schottke diodes was so silent it was quieter than my amp.  The Hexfreds vs the Schottke pieces are a toss up for me because your equipment is still quieter than the noise floor on my amp.

Until such time as I purchase a scope, I cannot say definitively if the Stealths are better than the Schottke pieces. 

From what my friends with scopes tell me is that Stealth Hexfreds are the way to fly....(I had to get back to that Boeing referece). 

Jim the overall price is just pennies more than the Schottke's.   

[hagtech]

Oh yeah, I would have to say these are ALL wrong for the heater application.  Great for B+ or a high frequency switching supply, but they're not right for heaters in a CORNET or CLARINET.  Waaay too much forward voltage drop.

jh

[jameshuls]

If I understand the data pages for the various diodes correctly, V_F (Instantaneous Forward Voltage/forward voltage drop) is the specified voltage drop for the diode. If this is the case, comparing them reveals the following:

Diodes Inc. 3A, 20V Schottky: 0.475V @ 3.0A @25^oC
                                          0.850V @ 9.4A @25^oC

Fairchild 8A, 1200V Stealth: 2.8-3.3V @ 25^oC @ 8A
FRED                                2.7-3.1V @ 125^o C @ 8A

IXYS HiPerFRED 15A 1200V: 1.79V @ 150^oC @ 15A
                                      2.75V @ 25^oC @ 15A

IXYS FRED 12A 1200V: 2.1V @ 150^oC @ 12A
                                2.7V @ 25^oC @ 12A

The warmer they are and the less amperage they receive, the lower the forward voltage drop in the FREDs. When graphed, the voltage drop increases exponentially with current, with a concave curve. So, if used at a small percentage of their rated amperage, the voltage drop should minimal, although its rise will be very quick. This means using a FRED rated for far greater amperage than necessary may resolve the voltage drop issue. I am hoping the 15A HiPerFREDS I will be using (along with the 4.3V coming off the transformer) won't cause an issue with low voltage for the H+. I can lose about 1.3 V.

Schottky's exhibit a very steep convex rise in forward voltage drop with amperage meaning very little inital change with the current, although with increasing temperature, they will become less efficient. I wonder if the Silicon Carbide Schottky's would improve performance? I have read many posts to the contrary, actually.

This temperature and amperage data highlights the dynamic performance stats of all these passive pieces in a circuit. Yet another variable/dimension the designer must consider! Not to mention the explanation it provides for change in character (usually for the better) of electronics with a warm-up, as I am sure every good designer would want peak performance at operating temperatures. For example, I find the Piccolo requires about 1/2 hour warm-up before an edge to the treble diminishes.

Whether this modification is overkill obviously depends on the resolution of your system. Tubes found that the change was below his audible noise floor.  I have no idea yet whether this will have any audible effect in my system. If I did, I probably wouldn't do it. It is a type of gambling, and the cost of losing (or breaking even) is negligible, or perhaps even positive -  a few bucks, a few hours and a few beers... and you can always go back. If you win on the other hand, I can think of no greater satisfaction. It's hard to find a reason not to do it!

If I end up with a component that can rival the very best, but can't yet appreciate due to other bottlenecks in the system, it seems, given the cost, to be a worthwhile investment. Put another way, if I have a component that I am quite sure is not going to need further modification, that is comforting to my obsessive need for perfection; I can direct my attentions elsewhere.

In fact, this is why, despite not knowing the difference between a resistor and a capacitor a year ago, I decided to build some Hagerman kits in the first place. I could see no other way of achieving the kind of performance I was interested in, short of putting down at least $5,000 per component. It  now has become an addiction. I now understand the term audioholic"". Whereas before I was just a music lover who wanted good sound, now I'm a music lover who wants good sound and who's obsessed with building my own components (facilitated of course by Hagtech) and modifying my existing system. It's comparable to lusting after a Ferrari, then realizing that for a tenth the cost you can buy the blueprints and build one yourself. Given the amount of forums dedicated to the topic, my situation is not unusual.

Anyways, I'm keeping the cornet stock until I have it up and running smoothly. Then I will start the mods and post accordingly.

[tubesforever]

From what Jim is saying here, I think the Stealth Hexfreds might not be the best choice for the H+ circuit.   

To take them out of my working units, I had to cut off the leads, apply enough heat for the lead to drop out of the PCB, clean the trace, heat shrink the Hexfred, bend the leads to work with the PCB, solder them in and then rework the H+ drop down resistor.  This ended up being about an hour of work per unit.  At least the Hexfreds I selected provided me with enough H+ voltage to adequately heat the tubes.  I am getting 6 volts and up in my application. 

While I would love to know if a scope would see any difference at all, I suspect as Jim mentions, that for the H+ DC voltage the Hexfred is not the optimum piece. 

James, I think you said the Schottke's were backordered by Digikey.  If that is the case then solder in the Hexfreds and see where you are sitting.  They will not create any mischief as long as you have adequate voltage at the H+ for your needs.  You might find it works out exactly as mine have.   I do not plan to remove mine since the unit sounds exactly the same to my ears and the high speed and soft recovery on these are superior to other diodes.

Jim's design work is excellent!

I wish my amps were as quiet as his Clarinet and Cornet2.

[jameshuls]

Actually Jim, I already have the Schottky's soldered in. The HEXFRED project won't happen for awhile; I am just purchasing the parts for now. Before anything I will put in the Obligatto PSU caps, then bypass the Janzten output caps with the FT-3 teflons. More important than all of this of course is getting the right R223 value and building the chassis - construction has started on the latter, and the order is in on the former.  aa

[Bill Epstein]

Been too busy working to run the Cornet much and, as all the parts are new, can't say how it sounds with the Kimbers and Jantzens.

I ordered 1N5821s to go back to stock from the Fairchilds from Allied http://www.alliedelec.com/Catalog/pf.aspx?FN=1773.pdf along with a Hammond 12x8x3 case for the Clarinet.

Happy Holidays to all.

[hagtech]

Oh yeah, you reminded me - there's a second reason for that 1.8 ohm or 2.2 ohm resistor.  It's not just for ripple low pass filtering.  The other function is tube warmup!

You see, when cold a filament's resistance is about 30% of operating nominal.  That means you get roughly 3x inrush current through the filament at turn-on.  If the power supply can deliver it.  This is one reason I hate linear voltage regulators for heaters, unless their current limit is carefully set.  What this does is cause tube failure via broken filaments.  A quick way to ruin a great tube.  The more you hammer the filamant with harsh turn-ons, the less it lasts.  Therefore, that 2 ohms of resistance becomes a current limiter, providing an inherent soft-start for the tubes.  It works splendidly.  You need not worry about power cycling a CORNET2 or CLARINET.  Or any Hagtech product.  We think of these things for you.

jh

[tubesforever]

Jim, I have two questions:  1. based on the assumption that the filaments get "shocked" during start up, would it be better to just let the equipment stay "on" all the time? 

If the heater remains at 6.0 to 6.1 volts would you think the tube would last longer than cycling it on and off with the Schottke diodes?

2.  with the hexfreds on fire up they start at about 5.9 volts and then creep up to 6.0 volts within 10 minutes.  I was hoping to hear from Bill if his hexfreds generate higher voltage after warm up. 

Before I used the hexfreds, with the Schottke's and the resistor in the circuit the voltage hit 6.0 volts and then rose up to 6.175 volts in less than 1/2 the time as the hexfred start up.   

Jim, the information about hexfreds claims that their soft recovery feature is even more important than their fast response times.  The claim is that this is one of the more important reasons to use them in tube DC environments.  Perhaps the Shottke's and the hexfreds both provide this feature. 

I guess I have more research to undertake. 

[hagtech]

1) On all the time only helps with certain equipment.  HAGTECH stuff does not have problems with turn-on.

1.5) It is not a matter of 6.1 or 6.0 volts.  It is a matter of 3 amps versus 1 amp.

2) Yes, soft recovery is a great feature.  The Schottkys do not have.  But there are other ways of dealing with the problem (see snubber article).

jh