I am using a string of 3 x 91V 5 watt zeners to ground, and the Zs are connected to a 3w 4.7k resistor to B+, which is 390v. The junction of the 4.7K and the zener string is connected to the G2 of a PP pair of el84s. I thought that with this arrangement i would have a G2 voltage of about 273 volts. On the amp the G2 voltage comes to 300. What is wrong in my circuit or design to cause the difference? Thanks for the help.
Franks advice is good.
24.9 mA would be the nominal current flow thru the Zener stack because of the (nominal) 3 × 91 → 273 volt drop, the 390 volt supply, leaving 117 V across the 4,700 Ω ballast resistor. Again, still nominally,
The poor little 3 watt 4.7 kΩ resistor tho' is pretty close to its dissipation limit. I'd “watch it carefully” in the confines of a vacuum valve chassis, where things definitely heat up.
Just saying,
GoatGuy ✓
24.9 mA would be the nominal current flow thru the Zener stack because of the (nominal) 3 × 91 → 273 volt drop, the 390 volt supply, leaving 117 V across the 4,700 Ω ballast resistor. Again, still nominally,
P = I⋅E
P = 24.9 mA × 91 V
P = 2270 mW per Zener
Since you have 5 watt devices, that's clearly within allowable tolerances. I'd bet one of the Zeners is out of spec by a fair amount (or actually that the whole stack of them, with uncharacteristic Murphy's Law bad luck, is composed of 3 devices each of which are somewhat at the high-end of their ±10% tolerance…)P = 24.9 mA × 91 V
P = 2270 mW per Zener
The poor little 3 watt 4.7 kΩ resistor tho' is pretty close to its dissipation limit. I'd “watch it carefully” in the confines of a vacuum valve chassis, where things definitely heat up.
P = I⋅E
P = 24.9 mA • (390 - 3 × 91 → 117) volts
P = 2,900 mW
Real close to 3 W. The simplest solution depends on whether you really need 24.9 mA sinking, or less. If less-is-OK, then just swapping it (4.7 kΩ) for a higher value solves the problem easily. Or putting an additional resistor in series, also solves the problem, by a different path. P = 24.9 mA • (390 - 3 × 91 → 117) volts
P = 2,900 mW
Just saying,
GoatGuy ✓
Being you are young, smart, good looking and so on (LOL), perhaps you can answer a long-standing question I have had in this topological conundrum.
With rarified gas regulator tubes, (e.g. 0D3, 175 V strike, 150 V nominal regulation), putting a capacitor in parallel is verboten, except at the smallest capacitor values to soak up HF/RF stochastic ionization Schott noise. Larger values drive the system toward nonlinear oscillatory dynamics and gas reg tube destruction.
So we don't.
Zeners seem so much more well behaved.
There is no strike-versus-run hysteresis per se, so no accumulated overcharge in a capacitor in parallel. Voltage rises, the VZ is attained, and with a spectacularly steep EI curve, the Zener breakdown conduction commences. Without hysteresis, it remains linear.
Is that the right interpretation as to why we CAN use modest-value caps in parallel to Zener stacks?
In the past, out of (essentially) fear of The Hysteresis Problem, I have protected each Zener and capacitor with a small value resistor in series with the capacitor (in parallel to the Zener). 1 µF cap, 1 kΩ R, 75 V ± 10% Z… and I've never had solid basis-theory as to whether the 1 kΩ is off by an order of magnitude or more.
Preëmptive thanks… GoatGuy ✓
The 4 kΩ will drive the Zener stack to higher quiescent current by +20%. This is not a problem for them, so that'd work. However, its own dissipation also rises by the same 20%. So, 2.9 W goes to 3.5 watts. This is well below the 5 watt dissipation, so it is definitely doable.
By comparison, the 7.5 kΩ resistor with 117 V residual nominal voltage drop then allows E = I⋅R … I = E/R → 117 ÷ 7.5 kΩ → 15.6 mA current flow, which is (15.6 mA ÷ 24.9 mA × 100) → 63% of, or 37% less, than the original 24.9 mA operating condition. If the designer had a reason for 25 mA, well … then 7.5 kΩ results in a current flow pretty far from the design spec. If however “25 mA” was pulled out of a tall black hat while drinking fine aged rum … then perhaps dropping to 15 mA with the 7.5 kΩ resistor isn't a problem.
So, the answer kind of depends on the rum.
LOL
GoatGuy ✓
Actually, this is for a stereo amp and both channels G2 measure within a few volts so that seems to "rule out" a bad zener. I have had in the past bad SS devices that caused problems which had me chasing my tail for a few days. thanks for the tip.
GG, the choice of the current dropping resistor was simply based on an estimate to keep the zeners from over dissipation. From my limited understanding the total current through the circuit will be the same and so erring on the conservative side, going to a larger resistor would place a smaller strain on the zeners.
I guess the only question would be whether there is any sonic effect. Both 5 watt resistors are the same type and manufacturer, which would make there sonic contribution the same.
As zener (or avalanche) voltage increases so does positive tempco. Below ruffly 5V the tempco is negative, becoming about -0.1% for the lowest values. So a 2V zener has ca -2mV/K tempco. As teh voltage goes up it becomes more and more positive. 50V and above the tempco is about 0.1%*V/K. That is a huge positive tempco. a 91V zener will increase it's voltage ca 90mV/K!
The power rating is at 25C. Flushing 25mA thru them heats them how much? Thats ca 2.2W! They are likely > 100C junction! So no wonder each zener is at about 100V.
Either live with the temp drift, it will eventually settle, but do you really need 25mA? I'd try 5...
Or series connect 18 5.1V zeners per 91V. That's only 54 in series, get some kid to hook em up or something. Or use some other regulation. Simple resistor divider, buffered by a transistor? Kinda like a cap multiplier?
I picked up a rule of thumb for a zener supply, years ago, although where from I dont know. Quiescent current draw through the zener should be 10 times more than the expected load current. So I figure if you have 2mA screens, then perhaps 20mA through the zener regulator should be about right for stiff regulation. Also anecdotally, when using 1N53** series diodes, to set 250V using 100V and 24V diodes, and a 100uF bypass, my screen supply takes a good 2 mins to stabilise and reach maximum regulated voltage. I'm also using a similar series resistance, and current flow. I now assume that the temp coeff. And temperature stabilisation of the zeners have probably more to do with this than the bypass capacitor. So I'm like to switch to 10 series 24V zeners, 50 series 5V is too much hard work! But also, zener screen regulation seems to work pretty good to me, better than a simple potential divider. A degenerated screen also works fairly well too.
Last edited:
Paraphrased for clarity.
OK.
Use the 7.5 kΩ then. No acoustic effect expected.
Remember, any load placed onto the Zener stabilized supply will NOT increase current thru this resistor, until the load exceeds the current nominally consumed by the Zeners! However, the VZ conduction voltage will vary a tiny amount, as is how the things work. Its why we like them generally.
Generally, contrary to oh-so-much of the popular über-audiophile literature, the 'sonic signature' difference between competent general purpose resistors and wickedly expensive tantalum-film-on-unicorn-hoof resistors, is nearly exactly diminimus ("zero"). Non-blind “golden ear” listeners profess to hear differences. My own triple-blind (unknowably randomized) testing shows exactly the opposite. Competent resistors have no 'sonic signature' themselves apart from whatever the effect is of have a particular resistor, in a particular location, in a particular circuit.
In other words, there very well could be an acoustic change by changing the value of the resistor, but it is not expected there would be a signature of for a given value, changing the manufacturer, so long as they make competent product.
Good luck.
GoatGuy ✓
A zener acts like a clamp in comparison to VR discharge tubes, right? While zener diodes are pretty constant as a voltage source, VR tube can deal with transient overloads, like in a screen regulator (as seen by the modulation of the music in the brightness of the tube), but they make noise. Shielding if necessary. Even a nixie as an indicator can make noise, proven by me touching or wrapping my hand around it. Zener diodes are noisy, but I think they are easier to "treat" with a simple filter. They just don't look as "cool" or whatever. If the ground is properly wired in the circuit, the noise is gone. In the case of the noisy nixie, it was indeed a bad ground connection.
Last edited:
I am not young, smart, or good looking. But I'll take a poke.
Gas tubes have a Change Of State from gas to plasma. This has a negative slope of resistance. In general, any negative resistance with a positive resistance has two stable states. With some combination of energy storage, neither state is stable, it bounces back and forth.
Zeners seem to have no zone of negative resistance.
Keep in mind to include the screen current in your calculations....especially during operation.... At idle it may be about 4 to 5mA ... At full power it can be around 15mA.... If you square wave, then you start seeing around 20mA ...this is per valve.... Look at the curve sheets and see what the screen current can be... Usually you dont use the Zener as the regulator in a power ciruit... You use the zener as the "reference" then use a pass tube to handle the load current...
With the 3 w resistor there is heat that i can feel emanating from the zeners and the resistors. That was a concern and so i started the post wanting to get some more insight as to what is going on and how close i am to a potentially disastorous event. But in terms of sound quality i am very happy with how the amp sounds this way.
So, since i want this amp to be trouble free i am going to use the 7K 5W resistor and that would result in less current flow through the zeners which would be a more reliable condition.
So, since i want this amp to be trouble free i am going to use the 7K 5W resistor and that would result in less current flow through the zeners which would be a more reliable condition.
+1.
If tube is too hastling (due to cathode-heater voltage ratings etc) a MOSFET or BJT will work.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.