Rocket rectifier tubes and Merlin's Rocket mod

[Mark]

Given the rectifier does have a hard time and the four notch Mullard seems to be the only rectifier capable of handling the 80F filter cap, why not look at changing the secondary winding so additional resistance can be added and hopefully make the rectifiers life a little easier?

http://www.valvewizard.co.uk/fullwave.html

Indigo colour text denotes info from the link above.

Edit: For example, when supplied by a transformer rated at 300-0-300Vrms, the EZ80 specifies Rlim(min) = 215 ohms per anode. The total impedance presented to the rectifier by the transformer is given by:

Rt= Rs + (n^2)* Rp
Where:
Rs = DC resistance of one half of the transformer secondary winding.)
Rp = DC resistance of the transformer primary winding.
n = Secondary to primary turns ratio (equal to the secondary voltage divided by the primary supply voltage).

If we were using a mains transformer with a 240V; 80R primary and a 310-0-310V; 50R per half secondary:
Rt= 50 + (1.29^2)* 80 = 183R
The EZ80 requires at least 215 ohms, so an additonal 215 - 183 = 32 ohm resistor must be placed in series with each anode (so we would probably use 33 or 47 ohms).
Because the limiting resistors will have to carry the ripple current of the reservoir capacitor it is best to use high wattage resistors. Even 7W resistors will usually get quite warm. Remember that the voltage drop across the limiting resistors will cause the HT fall proportionately.
Alternatively, a single limiting resistor could be placed between the transformer centre tap and ground, although its power dissipation will be doubled.

Moving on the the Rocket.

I have a Classic Tone Trainwreck compatible power transformer, the resistance between the centre tap to either end are 16 and 17 ohms respectively.

The primary resistance is 1.5 ohms for 120vac operation and 6 ohms for 240vac operation.

The turns ratio for 120vac operation is ([260X2]/120)=4.3333333

The turns ratio for 240vac operation is ([260X2]/240)=2.1666666

For 120vac operation Rt=17+([4.333333^2]X1.5)
=17+(18.7777777X1.5)
=45.166666ohms

For 240vac operation Rt=17+([2.1666666^2]X6)
=17+(4.69444444X6)
=45.166666ohms

Now time to look at valve specs.

First lets look at the common stuff.

http://www.drtube.com/datasheets/gz34s-jj2003.pdf

I may have it wrong here, but I'm looking at the capacitor input section at 2 X 300vac (Utreff), it states Rt = 2X75 ohms.

75-45.166666=29.833333 ohms

So a 33 ohm resistor per plate would assist the operation of this valve. I'm not sure but I'd guess- imate that the 33 ohm resistors would drop the rail voltage by about 10VDC.

Hopefully we can kiss the four notch rectifiers goodbye.

This is the discussion this thread came from.

http://ampgarage.com/forum/viewtopic.ph ... 0&start=15

[Tone Lover]

Good thoughts Mark I will use this in my next rocket build . also dont they state that if you use a diode before the rectifier it makes it able to use a larger cap. i will have to find that info and past a link to it.
Thanks Bill

[Firestorm]

Fender did something like this (according to HAD via someone else) resistors on the winding. Martin Manning recently did a sim that showed the problem to be hot switching. Indirect heaters (GZ34,GZ37, 5V4) tolerate heavy filters much better (because they come up slow).

[overtone]

Good to see more on Rt lim.
Going back over my notes from two Rockets, I had calculated 56R to add to each HT leg on "number one" and "number two" was only 3R called for.

But these were a couple of lifetimes ago and I cannot remember which was which anymore.
I think number one was Heyboer and two was with an old English PT, no wait, errrr. Well, won't be able to take a look inside until the end of the weekend.

[JoeCon]

After having a number of tube rectos burn out in my BB Reissue when I built my Rocket and Rockster I drank from Miles' Koolaid and said for $0.14 each silicon is the way to go. Simple and it just plain works...

[M Fowler]

After having a number of tube rectos burn out in my BB Reissue when I built my Rocket and Rockster I drank from Miles' Koolaid and said for $0.14 each silicon is the way to go. Simple and it just plain works...

Yes and you can always add a sag resistor on a switch if you feel you just have to have some SAG.

[JoeCon]

Mark

Is that the 2K2 3 watter in your 2 EL84 Rocket build??

[M Fowler]

Mark

Is that the 2K2 3 watter in your 2 EL84 Rocket build??

Yes it is.

[matt h]

(deleted)

[Tone Lover]

Yes and you can always add a sag resistor on a switch if you feel you just have to have some SAG.

Yes and you can always use some zeners to account for forward-drop in addition to the sag resistor and blow the minds of the "but they're not linear!" crowd.

... and UF diodes to cut down on line hash...

Do share how do you do that .
Thanks Bill

[matt h]

(deleted)

[Mark]

What concerns me with the Rocket is the rail voltage I really don't want to see it slip under 300 VDC, yet I want to give the rectifier every chance of surviving the 80uF filter cap.

The question I've asked myself about the 80uF cap is this. Does switching on cause the most damage to the rectifier or does constantly recharging the 80uF cap do the damage?

Once I've finished my Rocket I'll experiment with the R limit resistors. I'm hoping I have a 320VDC rail as I'll be able in implement the R limit resistors.

[Merlinb]

The question I've asked myself about the 80uF cap is this. Does switching on cause the most damage to the rectifier or does constantly recharging the 80uF cap do the damage?

Both.

Hot switching is *painful* to tube rectifiers, but maybe this amp doesn't have a standby switch, in which case this isn't an issue. ...Unless you happen to flip the power switch off-on fairly quickly, so the heater hasn't cooled down but the cap has discharged before you flip on again.... Solution? Don't flip you switches too quickly!

The max allowable reservoir capacitance is determined by the peak ripple current limit of the tube recto. The GZ34/5AR4 has a limit of 750mA peak, I believe. In fact, you can use any value of capacitance you like, *provided* you keep the ripple current below this value. You do that by making sure there is enough effective series resistance. Data sheets don't tell you this because often it is impractical; for example, using ten times more capacitance than the datasheet recommends usually means you have to add ten times more effective series resistance too, which means massive power wastage. No designer in the 1950s would want to do that.

Probably the easiest thing to do is to crack open a SPICE program, or Duncan's PSU designer. Model your circuit and take a look at the expected currents, and add series resistance until it drops below 750mA peak.

[Jackie Treehorn]

In my Dynaco ST-70, I was going through jj gz34's fairly quickly, arcing and such. That's four EL34's. What I did was put 2 11amp IXYS hexfreds in series with the rectifier. Solved all the problems and actually improved the sound of the amp a bit. Tighter, deeper in the lows. Doesn't change the voltages, either.

[martin manning]

Mark, there is an error in your calculation above. The turns ratio you used is 2x too big since only half of the secondary is in use in any given charging cycle. That means the Rt is only 23 ohms, and you would need to add 52 ohms to meet the requirement.

Jackie, it sounds like the issue with the JJ is inverse voltage, since the diodes aren't doing much of anything to limit current.

I'll run some numbers using SPICE and Mark's transformer data, and post results later.