Need help designing AC30 power supply with lower screen voltages

[Mike_J]

As we all know in the early 1960s when the first AC30s were made the wall voltage was lower than it is now.  An early Vox schematic shows the screen voltage at 320VDC.  Using a 280-0-280 volt secondary on a power transformer yielded about 332 volts on the first filter cap.  Given today's higher wall voltages, voltages of over 350 volts on the first cap and screen are likely if the original schematic for the power supply is followed. Furthermore, I have not seen a PT on today's market which would provide us with historic voltages using today's wall supply so a change of PT is not practical.
 
The AC30 uses EL84s which, according to the datasheet for the tube, have maximum high voltage and screen voltage ratings of 300VDC each. From what I have read it is not overly critical to keep the high voltage rating below 300 volts in order to ensure long-term power tube reliability.  A set of EL84s should last for many years with 350 volts at the first filter cap. However, I don't think 350 volts on an EL84 screen is a recipe for long life in an EL84. Thankfully, sluckey, PRR and others on the forum helped me find a reasonable solution for reducing the voltage to the screen supply using a VRR.


This is where this whole thing gets complicated. In order to drop 50 volts from the screen supply a mosfet will be used. Have heard mosfets prefer pulsating, unsmoothed voltage.  Not sure why. Maybe the pulses give it a break? Have also heard caps as high as 500uF or more can be used after a choke. So the question I have is where to place caps and at what rating?
 
Looking at where the voltage will be used might help. Voltage is converted from AC to DC at the rectifier tube and goes to the first filter cap.  That cap feeds the center tap on the primary side of the output transformer and sends voltage to each side of the primary feeding the power tube plates. Not sure how much filtering is needed at this stage? Maybe it is a good idea to lower the filtering here if the power tube plates don't need filtered voltage and it is preferable to have pulsating voltage on the mosfet then a smaller cap may be better? This node also feeds the choke.
 
A choke is in the circuit to block AC and pass DC forward if what I have read is correct. In the AC30 this is to the node that feeds the screen supply. Need to lower the voltage here approximately 50 volts. I have attached a schematic of the modified AC30 power supply for consideration by people more knowledgeable than I am. What I am trying to find out is what uF rating is suggested for the high voltage node? uF recommended for the screen supply? Where you would place the cap for the screen supply? Do you need a diode where I have it or anywhere else? How important is filtering to the power tube screen grids?
Thanks
Mike

[HotBluePlates]

This is where this whole thing gets complicated.

What if we go the dead-simple way?

As we all know in the early 1960s when the first AC30s were made the wall voltage was lower than it is now.  An early Vox schematic shows the screen voltage at 320VDC.  Using a 280-0-280 volt secondary on a power transformer yielded about 332 volts on the first filter cap.  Given today's higher wall voltages, voltages of over 350 volts on the first cap and screen are likely ...

332v / 1.414 = ~226vac.

What if you simply get a 120v:240v (or 120:220v) step up transformer, and use a solid-state bridge rectifier? The linked specs for Hammond's replacement AC-30 PT say the high voltage winding needs to supply 262mA, so look for a step-up isolation transformer with similar current capacity. It might be spec'd as (240v * 262mA =) ~63VA.

Now you skip the 5v winding (no tube rectifier needed) and just get a 2nd 6.3vac transformer rated for ~6.8A (again, according to the Hammond part spec).


Or... Contact Edcor to make you a custom PT with the specs you want.

[Mike_J]

332v / 1.414 = ~226vac.

What if you simply get a 120v:240v (or 120:220v) step up transformer, and use a solid-state bridge rectifier? The linked specs for Hammond's replacement AC-30 PT say the high voltage winding needs to supply 262mA, so look for a step-up isolation transformer with similar current capacity. It might be spec'd as (240v * 262mA =) ~63VA.

Now you skip the 5v winding (no tube rectifier needed) and just get a 2nd 6.3vac transformer rated for ~6.8A (again, according to the Hammond part spec).


Or... Contact Edcor to make you a custom PT with the specs you want.

I would take the custom PT route if it was my own build. However, this is for a friend who is a worship pastor. He travels around the country but only will use this amp in his studio. I wanted to make it quiet and reliable. He has had a lot of trouble with tubes blowing out. This is largely because the plate dissipation was over 16 watts and the already mentioned high screen voltage.


I am doing this for him no charge but don't want to buy a new PT considering he will almost never use the amp. This has been a learning platform for me which is why I am approaching it this way. Also, there appears to be an advantage according to many in keeping the plate voltages higher.


There is much logic in your suggestion otherwise. I have considered a custom made PT at least ten times just to be rid of this amp. I want it fixed right so I never ever ever ever see it again.


Thanks
Mike

[eleventeen]

Wouldn't the truly painless way to deal with this be to supply a variac and run the amp at 95%?


Especially if he's not going to move the amp. Just another box, I know. But it's a box that gets bought with money and dropped on the floor and it's done.

[HotBluePlates]

Or, if this is an already-built amp, just do the MOSFET regulator you drew to drop the voltage. You might have to adjust some resistor values on test to dial in the exact voltage you want.

... In order to drop 50 volts from the screen supply a mosfet will be used. Have heard mosfets prefer pulsating, unsmoothed voltage.  Not sure why. Maybe the pulses give it a break? ...

I've heard a lot of things too. Usually, they are out of context.

MOSFET's only have so-high voltage ratings and so-high power dissipation ratings. If you have to drop a LOT of voltage, pulsating d.c. means the power dissipation in the MOSFET is less than if the voltage (and the implied resulting current, which is more important here) were unwavering. You're looking to drop 50v for screen which maybe hit 20mA peak (if that), times 4 tubes, which is 80mA * 50v = 4 watts. Even if I guessed only half-right, you can waste that much heat in a MOSFET properly insulated/bolted to the chassis.

If you use the MOSFET to filter/feed the output tubes, then it has to be rated for essentially all the current draw of the amp, times the voltage you wish to drop, plus some sensible safety margin. You could do this in the sense of an "overall power scale/VVR scheme" but I'll defer to others who know more about these circuits.

[Mike_J]

Wouldn't the truly painless way to deal with this be to supply a variac and run the amp at 95%?


Especially if he's not going to move the amp. Just another box, I know. But it's a box that gets bought with money and dropped on the floor and it's done.

If my calculations are correct setting the variac at 93% would provide historic voltages and probably get him out of the woods.  Good suggestion. Will use if a better option can't be found. A new PT with a 260-0-260 winding would work as well. I am pretty sure if I was to build an EL84 based amp it would have a 260-0-260 PT.


Thanks
Mike

[Mike_J]

Or, if this is an already-built amp, just do the MOSFET regulator you drew to drop the voltage. You might have to adjust some resistor values on test to dial in the exact voltage you want.

... In order to drop 50 volts from the screen supply a mosfet will be used. Have heard mosfets prefer pulsating, unsmoothed voltage.  Not sure why. Maybe the pulses give it a break? ...

I've heard a lot of things too. Usually, they are out of context.

MOSFET's only have so-high voltage ratings and so-high power dissipation ratings. If you have to drop a LOT of voltage, pulsating d.c. means the power dissipation in the MOSFET is less than if the voltage (and the implied resulting current, which is more important here) were unwavering. You're looking to drop 50v for screen which maybe hit 20mA peak (if that), times 4 tubes, which is 80mA * 50v = 4 watts. Even if I guessed only half-right, you can waste that much heat in a MOSFET properly insulated/bolted to the chassis.

If you use the MOSFET to filter/feed the output tubes, then it has to be rated for essentially all the current draw of the amp, times the voltage you wish to drop, plus some sensible safety margin. You could do this in the sense of an "overall power scale/VVR scheme" but I'll defer to others who know more about these circuits.

Hopefully by using it only for the screen supply and only to reduce voltage 50 volts it will last a long time. I will make sure I install the mosfet properly.


Thanks
Mike

[2deaf]

Do you need a diode where I have it or anywhere else?

The 1N4007 diode should be on the other side of the capacitor right after the junction of the 10R resistor and the zener diode going toward the screens.

[Willabe]

I pulled out a couple of KOC's TUT books and this is how he shows the placement of the power scaling (PS) components.

He puts the 12v zener diode directly on the mosfet pins.

I trust his experience PS (VVR) circuits. 

[2deaf]

What is that red rectangle going to ground?

[Willabe]

What is that red rectangle going to ground?

Ground connection for the filter cap.

[2deaf]

So the red rectangle is a filter cap?

[Willabe]

Yes.

(I changed the drawing a little.)

[2deaf]

That's why you need the 1N4007 diode after the Source.  It keeps the positive DC from the filter cap. from hitting the Source when the gate voltage drops.  If there were no filter cap. after the Source, no diode would be necessary.

[Willabe]

There is a 1N4007 after the mosfet source.

[2deaf]

There is a 1N4007 after the mosfet source.

Yes, I see it and I told Mike earlier to position it where you have it.

I see how I chumped-up reply #13 which led to the reply that you gave me.  I should have quoted something from Mike so that it would be clear that I was addressing him. 

[2deaf]

I pulled out a couple of KOC's TUT books and this is how he shows the placement of the power scaling (PS) components.
He puts the 12v zener diode directly on the mosfet pins.

I sorta thought that the 10 ohm resistor between the Source and the zener came from TUT, but you show it without the 10R.  I'm not very sand savvy, so I don't know the function of the 10R resistor.  I tried leaving it out to see what would happen.  Nothing happened, so I didn't learn anything.  Does anyone know what the 10 ohm resistor does?

[Willabe]

Yes, I see it and I told Mike earlier to position it where you have it.

Yes you did, but I didn't understand you just now, I thought you were talking to me after I posted that hi lighted drawing. You had me scratching my head there. 


 

[Willabe]

I sorta thought that the 10 ohm resistor between the Source and the zener came from TUT,

Kevin shows it after the zener and 1N4007.   

[2deaf]

Kevin shows it after the zener and 1N4007.

Ah.  Now I'm wondering what it does at that position.

[Willabe]

IIRC Kevin was pretty strong on mounting that zener ON the mosfet pins, like mounting grid stopper R's as close as possible to the grid pin.

I'd have to re-read maybe 50 to 100 pages or so to find why he has the series R after the 1N4007.     

In TUT 4 that R is only 1 ohm.   

[DummyLoad]

one ohm or ten ohm make sense if you would want to measure current flow with a VM. a 100 ohms value would likely be inrush current limiting for the MOSFET protection under initial power up (e.g. caps are a short at turn-on), or a load short.  in this case, however, i assume that with the 1K between drain and power source that's covered.


PRR is the sand state expert: perhaps he'll enlighten the masses...   



--pete 

[Mike_J]

one ohm or ten ohm make sense if you would want to measure current flow with a VM. a 100 ohms value would likely be inrush current limiting for the MOSFET protection under initial power up (e.g. caps are a short at turn-on), or a load short.  in this case, however, i assume that with the 1K between drain and power source that's covered.


PRR is the sand state expert: perhaps he'll enlighten the masses...   



--pete

PRR told m to put the 1K resistor there. I followed up with a Mouser tech who said it was a good idea in order to keep high voltage rush from destroying the mosfet before the caps charge.  He said he had seen resistors as low as 10R to 1K in that position with 470R probably being the most common. PRR learned of this and showed some calculations I didn't understand in his reply. What I got out of the reply was you could probably use either a 470R of 1K resistor here. The 1K resistor will knock voltage down a little bit more than the 470R before it hits the mosfet so the mosfet doesn't have quite as much work to do which was PRRs reason for suggesting the value.


Thanks
Mike

[Mike_J]

Do you need a diode where I have it or anywhere else?

The 1N4007 diode should be on the other side of the capacitor right after the junction of the 10R resistor and the zener diode going toward the screens.

Thanks. I will make the change.


Thanks
Mike

[Mike_J]

I pulled out a couple of KOC's TUT books and this is how he shows the placement of the power scaling (PS) components.

He puts the 12v zener diode directly on the mosfet pins.

I trust his experience PS (VVR) circuits.

Thank you for posting that. I have a few of the TUT books but not that one. Knew it was important to have the 100K resistor as close as possible to the grid of the mosfet but I did not know about the importance of the zener placement.


I am in the process of correcting the schematic with 2deaf's and your comments included.  Also will post a layout. Still not sure what uF rating would be best for the caps.


Thanks
Mike

[2deaf]

I have learned the function of the 10R / zener network from another forum.  Once the current through the MOSFET gets high enough, the voltage drop across the 10R resistor in combination with the voltage difference between the Gate and the Source causes the zener to conduct and the current is limited to that point.     

[Mike_J]

Attached is the layout for the fixed voltage regulator screen supply.  I didn't know how large the circuit board should be so I just laid the components on the sheet of paper. In retrospect I should have started with a circuit board and resized it at the end. Live and learn, anyway the board size would be 3-1/8" X 4".


Thanks
Mike

[Mike_J]

Here is the schematic.


Thanks
Mike

[Willabe]

I have learned the function of the 10R / zener network from another forum.  Once the current through the MOSFET gets high enough, the voltage drop across the 10R resistor in combination with the voltage difference between the Gate and the Source causes the zener to conduct and the current is limited to that point.

Ah, it's just to kick in the zener to limit the current. 

[Willabe]

Here is the schematic.

That looks right to me.

I would try 20uF to 40uF for the screen filter cap. You show a 47uF in your layout, might be fine, but might be over kill? 

I think that the PS/VVR circuit does provide some dcv regulation so you can use a little smaller filter cap there?
 

[Willabe]

Knew it was important to have the 100K resistor as close as possible to the grid of the mosfet.......

On a mosfet it's called a gate not a grid.

[Mike_J]

Here is the schematic.

That looks right to me.

I would try 20uF to 40uF for the screen filter cap. You show a 47uF in your layout, might be fine, but might be over kill? 

I think that the PS/VVR circuit does provide some dcv regulation so you can use a little smaller filter cap there?

I used the 47uF because it was larger than a 20uF cap. If your layout will hold a 47uF cap it will easily hold a 20uf cap. However, I appreciate your advice concerning the cap sizes you would try. Vox used 16uFs on the screens so a 20 would certainly work. Only question is whether noise reduction would be helped with the larger cap. Will try both and see what happens.


Thank you for correcting the gate/grid confusion. As I recall when I first visited the mosfet it was to create a tubenit type cathode follower circuit using a mosfet, a 12V zener and other components as a replacement for a 12ax7 triode. If I remember correctly the gate did the same thing the grid does in a vacuum tube so I just think grid when I see a mosfet gate.


Thanks
Mike

[2deaf]

Here is the schematic.

I don't like it.
I have attached a possible alternate.

R3 (1K):  I don't think it's necessary, but I am probably in the minority opinion.  The R1/R2 voltage divider is not right if it follows a voltage drop that varies.  So I moved it to the plate node so that the screen voltage is always the same proportion of the plate voltage.

D1:  Doesn't look necessary to me, but if that is what TUT shows, go for it.  Ain't gonna hurt anything.

R5:  Needs to be in this position to function correctly.

C3:  47uf will hold the screen voltage up as it discharges when the plate voltage drops.   

[DummyLoad]

I have learned the function of the 10R / zener network from another forum.  Once the current through the MOSFET gets high enough, the voltage drop across the 10R resistor in combination with the voltage difference between the Gate and the Source causes the zener to conduct and the current is limited to that point.   

OK, i'll buy that. but in this case as shown, the resistor is outside of the gate-source ckt. so that's not accomplishing the task as described: move 10 ohm resistor between MOSFET source and zener anode. 


see attached.


--pete

[DummyLoad]

sim snapshot. i assumed the target B+ is 300V for G2 - if not adjust R2 value accordingly.


--pete

[sluckey]

Let me throw out a couple alternatives that may interest you...

Just use a NTE5209A 50v 10w zener. Can be set up to drop only the screen voltage or can be set up to drop screens and all downstream nodes by 50 volts. One component, maybe two if you want to count the additional filter cap.

Maybe a simpler tube circuit equivalent of the MOSFET circuit. Take a look at the schematic in reply #5 of Timbo's post. The 150K---220K voltage divider determines the voltage at node B.

     http://el34world.com/Forum/index.php?topic=19850.msg207964#msg207964

[Mike_J]

sim snapshot. i assumed the target B+ is 300V for G2 - if not adjust R2 value accordingly.


--pete

That is so much better than average.


Thanks
Mike

[kagliostro]

In addition to the previous solutions (the FVR or the Zener or the Tube Regoulator)

put on the amp 6P14P-EV tubes instead of el84

http://www.tubes.ru/techinfo/HiFiAudio/6p14pev.html

Franco

[Mike_J]

Here is the schematic.

I don't like it.
I have attached a possible alternate.

R3 (1K):  I don't think it's necessary, but I am probably in the minority opinion.  The R1/R2 voltage divider is not right if it follows a voltage drop that varies.  So I moved it to the plate node so that the screen voltage is always the same proportion of the plate voltage.

D1:  Doesn't look necessary to me, but if that is what TUT shows, go for it.  Ain't gonna hurt anything.

R5:  Needs to be in this position to function correctly.

C3:  47uf will hold the screen voltage up as it discharges when the plate voltage drops.

Regarding:
R3: I am quite certain it can be considerably smaller but I think it is a good idea to protect the mosfet drain at start-up until the caps charge. I appreciate the placement of the power source before the choke and not after it. Your pretty sure it is better to insert the voltage from this point?
D1: They cost almost nothing. Why not?
R5: Per Willabe's post #8 KOC placed the 12V zener diode directly on the mosfet pins. This can only be accomplished if the source side of the zener is moved to the left side of the 10R/5W resistor. Accordingly, moving it to the other side of the resistor as your schematic shows would not be KOC. Are you pretty confident about the way you have the resistor shown on your schematic? KOC is not God but he does sell a lot of books.
C3: You think 1uF is preferable to 47uF because holding the screen up as it discharges is a bad thing? I am very ignorant where this matter is concerned. However, if you take the voltage from the plates (plate side of the choke) then you are getting your voltage with very little filtering I would think. I asked how much filtering is necessary at the screen supply level because I don't know. Pretty sure it is very important by the time it gets to the preamp level but don't know about the screens.


Thanks
Mike

[Mike_J]

Let me throw out a couple alternatives that may interest you...

Just use a NTE5209A 50v 10w zener. Can be set up to drop only the screen voltage or can be set up to drop screens and all downstream nodes by 50 volts. One component, maybe two if you want to count the additional filter cap.

Maybe a simpler tube circuit equivalent of the MOSFET circuit. Take a look at the schematic in reply #5 of Timbo's post. The 150K---220K voltage divider determines the voltage at node B.

     http://el34world.com/Forum/index.php?topic=19850.msg207964#msg207964

Only want to drop the screen voltage. Where would the voltage from the zener come from. For example would you draw it before or after the choke?


Timbo's schematic is based on what HBP mentioned in an earlier post I think. Would be good for a start-up build but this is an existing amp that already has a PT that is sending voltage that is too high for current production tubes and wall voltages. Just want to fix that problem by getting the screen voltages down to a level the tubes can reliably handle.


Thanks
Mike

[sluckey]

I've attached a pic of the zener. It works on paper. But I've never done this in an amp.

I thought your ultimate goal was to extend the life of the EL84s? 350v or even 375v is not dangerous to the tubes. High plate current (or cathode current) is what shortens the tube life. Lowering the screen voltage is one way to lower the current through the tube. But, simply increasing the value of the cathode resistor will lower the tube current and give longer life too. I would try a 75Ω or even a 100Ω cathode resistor. That's probably all you really need to do.
 

[Mike_J]

I have looked at every schematic I could find related to the Dana Hall VVR.  In all of them the 10R/5W resistor is between the source on the mosfet and the 12V zener.  I am pretty sure 2deaf's response concerning R5 is sound so I am going to switch it back. Think the safe thing to do is to make the power cap at the output of the circuit 20uF for now as per DummyLoad's schematic. Still not sure about whether it is best to pull voltage before or after the choke. That is still confusing to me.


Thanks
Mike

[DummyLoad]

screen PS tap after the choke.


--pete

[2deaf]

R3: I am quite certain it can be considerably smaller but I think it is a good idea to protect the mosfet drain at start-up until the caps charge. I appreciate the placement of the power source before the choke and not after it. Your pretty sure it is better to insert the voltage from this point?

I changed it so that the power to the screens and the reference voltage both take off after the choke.  That way the reference voltage has two filter stages.

R5: Per Willabe's post #8 KOC placed the 12V zener diode directly on the mosfet pins. This can only be accomplished if the source side of the zener is moved to the left side of the 10R/5W resistor. Accordingly, moving it to the other side of the resistor as your schematic shows would not be KOC. Are you pretty confident about the way you have the resistor shown on your schematic? KOC is not God but he does sell a lot of books.

I don't have the TUT book so I don't know exactly what he diagramed.  He may have been leaving the current limiter out and the resistor after the Source/zener was for some other purpose.  If you buy into the current limiter idea, the 10R resistor will have to be between the Source and the zener.

C3: You think 1uF is preferable to 47uF because holding the screen up as it discharges is a bad thing? I am very ignorant where this matter is concerned. However, if you take the voltage from the plates (plate side of the choke) then you are getting your voltage with very little filtering I would think.

The power for the screens comes from the screen side of the choke giving it two filter stages.  I changed my schematic so that the reference voltage is also taken from the screen side of the choke.
If you play a power chord and the plate voltage drops ten percent, there will be a lag before the screens drop ten percent because the 47uf cap. needs to discharge to that level.  Maybe no big deal.   

[Mike_J]

Attached is the latest layout diagram based on 2deaf's schematic on the previous post. Additional comments I have are if I had a 200R/5W resistor and no 1K/5W I would use it. Should be sufficient. Probably will experiment with electrolytic cap from no cap at all to maybe as high as 220uF just to see what difference it makes.


Thanks
Mike