In the photo above, I have a 7824 Regulator. It's max voltage is 40V (or something like that.)

Here's my theory: There's 340V coming in, but only 40V is dropped across the IC, and 300V across the zener diodes.

Am I correct here? Is it okay to assume that the IC would only "see" 40V?

Thanks,
Josh

Yes but ....

If you are using the standard TO-220 uninsulated package, the bare metal back is connected to the Com terminal. Which means if you are using a heatsink, it will be at + 300 VDC with respect to the bottom of your diode string. Since you are dropping approximately 16 Volts across the regulator, a heatsink will be most likely required.

IMHO, it is not most reliable design to operate the regulator up against it highest voltage limit. I personally would use a zenier diode string and dropping resistor to get the voltage to 330V and then use your circuit with a LM7806 to regulate it to 324 Volts. That will give you a little more protection against high voltage input.

Larry C

I concur with Larry, and I have some other concerns.

If there's any capacitive load at all, then if the input voltage rises sharply, the 78xx will see the full voltage momentarily. Also, without other protection diodes, if the input voltage suddenly drops to zero (e.g., accidental input short), it sees high reverse voltage.

If the output is momentarily shorted, you get the full differential across the 78xx.

If the zener voltage is on the low end of its tolerance and the input voltage is high, you exceed the 78xx rating.

In any of these cases, you have a dead chip.

With a little more info, I'm sure there are quite a few folks here who could suggest a better alternative. (Min/max input voltage, min/max load current, load type?)

And still more concerns...
Regulators like these give you a spec for maximum input/output differential voltage. The 7824 is a little higher than the rsst of the 78XX series. But here, you have to think more along the lines of the Vceo rating of the output transistor stage. If you are charging load capacitance, there would be close to 350 volts across the output stage. Another thing to remember is that diodes are noisy creatures that should be bypassed when establishing a reference. also, the "com" or "adj" pins of regulators tend to be high impedance inputs. Consider using a biasing resistor between 1 & 2 to set the current for the best regulation. Those 3 zeners are probably really 6 diodes internally and I doubt that you will be that close to 300 volts unless you have many diodes to choose from.

I guess I need to clarify. Did you mean a "7824" or are you using a newer "78024"? Keep in mind the old trick of using a resistor between the common and output to establish a fixed reference current. A resistance is then placed between common and the negative supply rail to drop the desired additional voltage. Bypass capacitance is required here just as it would be with your zeners. Then you still need to add the protective diodes.

Just as John mentioned, it would be interesting to know exactly what you need with this supply. I was playing with a PWM supply concept using a 555 driving a darlington opto-coupler in the biasing circuit of the series pass transistor. I was going to see if I could get decent regulation just working with the control voltage. It wasn't something that I really needed though, so it got back burnered.
Joe

OK, thought about it some more. Add a high wattage zener, cathode to the left, between input & output. Choose 1 that is greater than the steady state voltage drop across the regulator when things stabilize but less than the max voltage difference that the regulator can handle. Add a protection diode between common and output, cathode up, to discharge your reference filtering capacitance. Use at least a 1N4004 or equiv. Add a biasing resistor between your input and zener string and capacitance across the zeners. If your source is hi-z, you might be fine. If lo-z, consider a series resistance before the regulator that gets bypassed once your output gets close to the desired value.
Joe

[This message has been edited by JoeTestingEngr (edited 10-02-2006).]

As far as what I need the circuit for, I'm building a tube amp. The power amp section is staying the same, but the preamp section is being changed out, and it needs 325V instead of the 260V provided. I don't know enough about RC and LC filters to redo the power supply like that, and I figured it's easier to use a filter cap on a clipped sine wave than a reduced sine wave.

What I have shown here is just an example to make things easier. Also, the 78024 should be 7824. I'm using the TO-216 package. It's all plastic with a mounting hole for a heat sink, so the sink will be at 0V.

I was either gonna go it this way, or use an NPN transistor, Collector as input, emitter as output, resistor going from C to B, and zener diodes from B/resistor to 0V. The only problem is I forget how to make the feedback amplifier and what values to use for it.

It would be so much easier if they made a regualtor from the 7800 series that goes for high voltage, but this was the closest I could come up with.

OK Josh, is this just a line level pre-amp or more of a serious driver? How much of a resistive load do you think it will look like at min. and max. drive conditions? Is it OK for it to build up over a few seconds or will that hurt your other biasing needs? Is your 340 volt DC source from a full wave bridge and a filter cap unloaded? Or lightly loaded?
Joe

[This message has been edited by JoeTestingEngr (edited 10-04-2006).]

Do you even need a fully regulated supply for this application?

If this is a run-of-the-mill pre-amp the tubes will all be biased class A, so the average load will remain pretty constant anyway.

Joe and Paul - I'm not really sure, I downloaded some schematics and I'm copying them over. The preamp is for guitar to line level. The voltage is specified on the schematic. As far as how well regulated it needs to be, I'm not sure, but like I said before, I'm horrible at figuring out RC and LC power supplies like the ones in tube amps, that may have 4 or 5 voltages.

All I know is that the preamp calls for 325V. I actually have 450V coming in, but I was either gonna cascade a few of these so that the voltage drop + output voltage = less than maximum voltage for the device. The more I think about it, I'm better off with a bipolar transistor/resistor/zerenr ladder described in my last post.
like this:
http://en.wikipedia.org/wiki/Linear_regulator

Maybe you want to try this:

http://www.geocities.com/tjacodesign/300vsup/300vsup.html

(found by looking thru http://www.discovercircuits.com/P/pwr-hivolt.htm )

Just make sure the FET of your choice can stand the voltage and the power dissipation...

In my experience, tube circuits are very tolerant of voltage swings, and if you use a big enough (47+ uF) electrolytic condensator for the hi voltage supply, the small current drawn by the preamp will be regulated enough.
One similar, simple, tried-and-tested solution is two 47uF, 350V caps between + and ground, with an 1 kOhm in series between them. The resistor will limit the voltage swings from the rectified AC flowing into the second cap.
Oh, and when testing these circuits: Do not forget to add a load ("bleeder resistor") at the output or you might be fully awake in a spilt second ;-)

edited to fix link

[This message has been edited by StarTrek (edited 10-05-2006).]

Josh, I don't think you need to regulate at all. The class-A preamp stage will draw a fairly constant current, and the voltage is not especially critical. Just select a series resistor to drop the right amount of voltage and add an electrolytic filter cap after it for hum reduction, similar to StarTrek's suggestion.

The electrolytic should be rated at the full supply voltage, however, because there won't be any voltage drop until the preamp tube warms up and starts drawing current. Alternatively, lower the values of the voltage-drop and bleeder resistors to form a "stiffer" voltage divider that will limit the cap voltage during warm-up. Watch the power ratings on those resistors.

Note that the circuit referenced by StarTrek is not regulated... It just exhibits a lower output impedance than a simple voltage divider, and is variable.

Glass audio is a great hobby for the Winter months!

[edit: Oops, wrong UBB code]

[This message has been edited by John Crighton (edited 10-05-2006).]

That's what I'd go with too, but just watch the peak voltage on the capacitors to allow for the switch-on surge before the pre-amp tubes warm up and start to draw current.

Does the power supply use an indirectly-heated rectifier for the B+ supply by the way? If so, then the peak voltage might not be that much higher anyway because the rectifier cathode needs to warm up just like the preamp tubes before B+ will appear.

It's much more of a problem if you're using semiconductor rectifiers so that the full B+ comes on before any tubes are drawing current to "pull" it down.

P.S. Do you know the configuration/tubes used in the preamp? 325V sounds a little on the high side for a small audio preamp.


[This message has been edited by pauluk (edited 10-06-2006).]

Thanks to JoeTestingEngr for the following:

This is where you tube gurus can tell me if this concept works. The zener would probably be just under the max differential voltage. (35V???). The top pot would be a high R, just to set the biasing for the control grid for optimal performance. 24V/Rtop sets the ref. current(Iref). (Vmax-24)/Iref=(R fixed + R adj.) The output cap is there for stability and should be >= 1uF. The adjustment filtering cap might be more than 1 cap. If you use a large capacitance for a slower ramp-up, you would probably use a small disk for RF bypassing. The zener should double as the I/O protection diode and the diode will discharge the filter capacitance.

This circuit puts the adjustment pot closest to circuit ground. It works with the assumption that you would want a given high voltage +/- a certain amount. You would take your voltage range / I ref to determine the closest linear pot value and make up the rest of the total with the fixed resistor. You could also use a multi-turn pot and no fixed R or a low R "FINE" adjustment in series with a high R "COARSE" adjustment.

So this is where Paul and John and Mike should chime in with triode suggestions (cuz I'm clueless!) The last tubes I worked with were 1KW broadcast tetrodes and 55KW klystrons, and I'm powerless here. I'll vouch for the part past the tube. Even if this circuit doesn't fit your pre-amp, it wouldn't hurt to have it around to tweak your other tube circuits.

As always, HAVE FUN!!!
Joe

Joe, I agree with most points, but not with the "Iref" equations. Iref is indeed the current through the reference divider, but it's supplied through the 7824 pass element, not through Rtop.

For stability, Iref is normally selected to be at least an order of magnitude above I(Q), the quiescent current sourced by the IC through the COM pin. For the 7824, typical I(Q) is 5 mA, so Iref would be set at 50 mA or so. The current supplied through Rtop is something like 1 mA, and the rest is supplied through the tube cathode and the IC.

However, that 50 mA Iref is a lot of wasted power for a 10 mA load, so a different regulator exhibiting a lower I(Q) would be better. Consider the LM317, which is essentially a "7801.25" (1.25 V regulator) having a maximum I(Q) of only 100 uA. You'd then select Iref to meet the LM317's minimum-load requirement (12 mA max), which is conveniently two orders of magnitude above I(Q).

You could probably choose reasonable fixed values intead of using a trimpot for Rtop. Consider how this circuit works: The OUT pin of the regulator IC is always negative with respect to the IN pin, and it's this negative differential that operates the control grid. The IC compensates for greater loads by increasing the conductance between its IN and OUT pins, making the control grid voltage less negative and raising the tube's conductance.

The problem is that there's a limit to how low the IC can make its IN-OUT voltage; for the 7824 it's about 2 V, and for the LM317 it's about 3 V. This voltage applied negatively to the control grid may not be low enough to allow the tube to supply the required full-load current, so the Rtop divider is in place to reduce it. However, the divider ratio must not be so low that the tube can't be controlled at the minimum-load condition when the IN-OUT differential is greatest (limited by the zener voltage).

This is probably a lot more than you wanted to know... but it's part of HAVING FUN, right?

John, I regretted not using component numbers after I had zipped the email off to Paul. I was referring to the top fixed resistor as Rtop since the pot was a VR. I agree with you completely because I was thinking of the LM317 or 350 all along. But Josh already mentioned the 7824. If I had known it was 5mA,(Are you sure?), I wouldn't have mentioned it. PDs of the reference string resistors start making it more of a PS/Heater. My datasheet didn't show specific values for the 7824. Not knowing the biasing requirements for a given tube, I thought tying to the Regl output might be pinching off too much. I was wondering if a little R in parallel with the triode might not be good as well.
So do we agree that with the LM317 or LM350 and a suitable triode, Josh might pull it off? I haven't read any tube data in over a decade. What about the possibility of a little R between the cathode and regulator input to set I limit and help with grid biasing?
Joe

[This message has been edited by JoeTestingEngr (edited 10-10-2006).]

Joe, I think it would work fine using an LM317 without any further tweaking. (Yep, the Fairchild LM7824 data sheet says I(Q) is 5 mA typ, 8 mA max.)

Since there are so many ways to skin this particular cat, I don't think I'd bother using a triode for this job. I was thinking along the lines of a TL431 with a cascode NPN driving an N-channel series-pass MOSFET. Guess what? Somebody else beat me to it .

Happy soldering,
John

Wow, all those years of reading Japlish in Sony Broadcast equipment manuals hasn't prepared me for Russlish. C'mon, was that really how you were going to design it? I was thinking along the lines of a single supply jfet op-amp with a zener bleeder supply and a bipolar series pass. Just a protected voltage divider input and reference scheme. Should we have just considered the ye ol' LM723 ?

Although the text is not always crystal clear I think the schematic speaks for itself: It's nicely simple and effective, and since the only IC is in a TO-92 package, it's also dead-simple to breadboard.

Of course, in a world where even the AC power cord has to be specially braided, rhodium-plated neutrogen-free copper with argon-filled polygooglene foamed insulation for impedance matching of sub-nanosecond microphonic transients, we all know that semiconductors of any kind are evil. The only proper way to do this is with a separate ferroresonant transformer for line regulation and a mercury-vapor rectifier.

Coincidentally, one of my current projects is a small production run of replacement power supplies for the military. Since it was designed in 1975, of course it uses... LM723!