High Voltage Regulators / Reducers

[andrewRneumann]

Alright folks, I've come a plateau in my knowledge. I really need some guidance on this one. I'm not a solid state guy. Aside from diodes and Zener diodes, I don't know Peter from Paul or even Mary. But I think I'm going to have to dive in and figure some of this out. Here's my situation...

I have a B+ supply of 400VDC. I want to run the power tube screens (and some other equipment) at approximately 250V. The issue is that I need some regulation of the 250V over a pretty large change in current... I'm estimating 10mA at idle and about 50mA at full drive. I don't need perfect regulation, but I need to keep it above 200V. Obviously a 15K dropping resistor is NOT going to work.

I understand shunt regulation with Zeners. The issue with that is that it's wasting a whole lot of power. If my back of napkin estimates are right, at idle I would be dumping 19W out the window through a large current limiting resistor and a string of 5W Zeners. Doesn't seem like the right answer.

What is the right answer? My reading seems to indicate that a MOSFET is the best solution. Merlin has an interesting voltage follower that I could set up between B+1 and B+2. It utilizes a MOSFET and a BJT (for current protection?). I could copy and paste this design, but I have no idea how to find the right MOSFET and BJT for my needs. Can someone hold my hand and lead me in the right direction?

 

[peteb]

I have a B+ supply of 400VDC. I want to run the power tube screens (and some other equipment) at approximately 250V. The issue is that I need some regulation of the 250V over a pretty large change in current... I'm estimating 10mA at idle and about 50mA at full drive. I don't need perfect regulation, but I need to keep it above 200V. Obviously a 15K dropping resistor is NOT going to work.

what do you want such a low screen voltage?

if you set the screen to idle at 250 VDC, what is going to cause it to drop to 200 or below?

the screen current is a small fraction of the plate current. Why would there be large changes in current?

 

[Ten Over]

What is the right answer? My reading seems to indicate that a MOSFET is the best solution. Merlin has an interesting voltage follower that I could set up between B+1 and B+2. It utilizes a MOSFET and a BJT (for current protection?). I could copy and paste this design, but I have no idea how to find the right MOSFET and BJT for my needs. Can someone hold my hand and lead me in the right direction?

I might be able to help you with the components if you post Merlin's schematic.

Otherwise...

 

[Ten Over]

Here is one with over-current protection:

 

[Jon Snell]

A schematic of your PSU will help.
Most manufacturers take a 50% tapping from the bridge rectifier to supply medium voltage components.
Using a series regulator is very inefficient but may be the only way depending on your psu options.

 

[andrewRneumann]

Here is one with over-current protection:
View attachment 984822

Thank you. Can you walk me through this data sheet and make sure I’m following the rules? This is the first MOSFET listed on your example schematic.

Vdss is 900V. I‘m using a PT with Vpeak of 470V so there’s no way I can exceed that downstream of the rectifier. Not using a choke so no flyback voltage. Check.

Id is 2.8A at 100C. My estimated max average current is 50mA. Inrush current should be limited to 10 times that… so 500mA. Check.

Pd is 1.64/C … say 164W. That’s a lot. I need only 150V x .05A = 7.5W (times a healthy safety factor of 3 would be 22.5W). Check.

Have I covered the bases correctly? What else is important on the data sheet I need to plan for?

Finally, how do I know I can properly dissipate the heat?

 

[andrewRneumann]

A schematic of your PSU will help.
Most manufacturers take a 50% tapping from the bridge rectifier to supply medium voltage components.
Using a series regulator is very inefficient but may be the only way depending on your psu options.

Yeah it’s just the typical tube amplifier PT with a CT HT and 50v bias winding. If there was a PT out there that had the proper tappings for taking off the voltage I need, I’d sure love to know about it. Do you have any suggestions?

 

[printer2]

Yeah it’s just the typical tube amplifier PT with a CT HT and 50v bias winding. If there was a PT out there that had the proper tappings for taking off the voltage I need, I’d sure love to know about it. Do you have any suggestions?

Use a center tap transformer where the full winding will give you your 400V with a bridge rectifier. So a 160V - 0 - 160V or there about transformer.

 

[Ten Over]

Thank you. Can you walk me through this data sheet and make sure I’m following the rules? This is the first MOSFET listed on your example schematic.

View attachment 984949

Vdss is 900V. I‘m using a PT with Vpeak of 470V so there’s no way I can exceed that downstream of the rectifier. Not using a choke so no flyback voltage. Check.

Id is 2.8A at 100C. My estimated max average current is 50mA. Inrush current should be limited to 10 times that… so 500mA. Check.

Pd is 1.64/C … say 164W. That’s a lot. I need only 150V x .05A = 7.5W (times a healthy safety factor of 3 would be 22.5W). Check.

Have I covered the bases correctly? What else is important on the data sheet I need to plan for?

Finally, how do I know I can properly dissipate the heat?

The one that I listed has the TO-220F (aka TO-220FP) package which is sometimes called "Full Pack". This is the one that has an "F" in the middle of the product number. I like this package because I can bolt it directly to the heat sink without having to fool with that little plastic thermal wafer. It doesn't dissipate heat as well as a regular TO-220 package, but it is available with a dissipation that will work in this application.

The spec's sound like radical overkill, but just as always, no two of them can occur at the same time. It can handle 900V and 8.0A, but not at the same time. The device is limited by its wattage rating and the wattage derating with temperature. In this case it is 68W at 25 degrees and that wattage derates at 0.54W for every degree hotter than 25. So if we took a worst case Tc of 100, the power would derate by 75 degrees x 0.54W = 40.5W leaving us with a limit of 68W-40.5W=27.5W. It doesn't sound like radical overkill now, especially since you were looking for a minimum of 22.5W.

The Safe Operating Area is always useful when selecting these things.


You mentioned 50mA, so I doubled it for 100mA because the current limiter would probably limit around twice the working current. The Drain-to-Source voltage is 150V and a point was marked on the graph at 100mA, 150V. I then drew a line for 27.5W which is the wattage at Tc=100 degrees. The point is well below the line, so I considered this device acceptable.

I just bolt these things to the chassis for heat sinking. It seems unlikely that the chassis would exceed 100 degrees Celsius.

 

[andrewRneumann]

The one that I listed has the TO-220F (aka TO-220FP) package which is sometimes called "Full Pack". This is the one that has an "F" in the middle of the product number. I like this package because I can bolt it directly to the heat sink without having to fool with that little plastic thermal wafer. It doesn't dissipate heat as well as a regular TO-220 package, but it is available with a dissipation that will work in this application.

The spec's sound like radical overkill, but just as always, no two of them can occur at the same time. It can handle 900V and 8.0A, but not at the same time. The device is limited by its wattage rating and the wattage derating with temperature. In this case it is 68W at 25 degrees and that wattage derates at 0.54W for every degree hotter than 25. So if we took a worst case Tc of 100, the power would derate by 75 degrees x 0.54W = 40.5W leaving us with a limit of 68W-40.5W=27.5W. It doesn't sound like radical overkill now, especially since you were looking for a minimum of 22.5W.

The Safe Operating Area is always useful when selecting these things.
View attachment 985066

You mentioned 50mA, so I doubled it for 100mA because the current limiter would probably limit around twice the working current. The Drain-to-Source voltage is 150V and a point was marked on the graph at 100mA, 150V. I then drew a line for 27.5W which is the wattage at Tc=100 degrees. The point is well below the line, so I considered this device acceptable.

I just bolt these things to the chassis for heat sinking. It seems unlikely that the chassis would exceed 100 degrees Celsius.

This is really helpful thanks. I know just enough to roast some silicon! I've already burned up a bunch of Zeners, resistors and fuses fooling around with my power supply. HT fuses are a very good thing. I've got a lot of reading to do on transistors.

Can you post a photo of how you physically mount and wire up these devices?

 

[Ten Over]

Can you post a photo of how you physically mount and wire up these devices?

I sanded the paint off and then used fine sandpaper to make a smooth surface so that the device makes a good thermal connection.


This one is an IGBT with one of those thermal wafers because it isn't a Full Pack package. The Drain on a lot of power MOSFETs and the Collector on power IGBTs is connected to the bare metal part that bolts to the heatsink. That means that the bare metal part has HT on it when used in a VVR. The wafer is a electrical insulator and a thermal conductor. Sometimes the hole through the bare metal Drain is out in the open so that you have to use the wafer and a non-conducting bolt or other arrangement to electrically isolate the Drain. One time I was running an experiment and didn't have a nylon bolt and didn't want to go into town to get one, so I wrapped a steel screw in pipe thread tape (I learned that one from Sovtek). It was pretty exciting for the second or two it took me to turn the power switch off.

The 470R resistor is soldered directly to the Gate pin. This resistor uses the capacitance of the device to form a low-pass filter that eliminates high frequency oscillation just like we do with tubes. I double heatsink the pins when I solder. The nut has some sort of thread lock on it.

 

[NTC]

It looks like you have your solution. If you decided you need a tube to do this job, check out a Gibson GA-86. It uses a 6V6 as a linear regulator to get 340V screens from a 415V source. The Gibson GA-100 bass amp uses an 0C2 regulator, though I think those have a fuxed voltage drop.

I have a Newcomb Audio KX-30 PA amp on the shelf waiting for some future free time that has an adjustable screen voltage. It also uses a 6V6 regulator. Radiomuseum.org has a schematic. https://www.radiomuseum.org/r/newcomb_kx_30.html

 

[andrewRneumann]

I sanded the paint off and then used fine sandpaper to make a smooth surface so that the device makes a good thermal connection.
View attachment 985518

This one is an IGBT with one of those thermal wafers because it isn't a Full Pack package. The Drain on a lot of power MOSFETs and the Collector on power IGBTs is connected to the bare metal part that bolts to the heatsink. That means that the bare metal part has HT on it when used in a VVR. The wafer is a electrical insulator and a thermal conductor. Sometimes the hole through the bare metal Drain is out in the open so that you have to use the wafer and a non-conducting bolt or other arrangement to electrically isolate the Drain. One time I was running an experiment and didn't have a nylon bolt and didn't want to go into town to get one, so I wrapped a steel screw in pipe thread tape (I learned that one from Sovtek). It was pretty exciting for the second or two it took me to turn the power switch off.

The 470R resistor is soldered directly to the Gate pin. This resistor uses the capacitance of the device to form a low-pass filter that eliminates high frequency oscillation just like we do with tubes. I double heatsink the pins when I solder. The nut has some sort of thread lock on it.

Interesting! Yeah I just read about how the attach point is part of the circuit. That was a big surprise. I'm leery of those shouldered sleeves that insulate the nut/bolt from the component. I just imagine that plastic getting brittle and breaking down over time and then poof. So what are the pros and cons of using an FP device with thermal compound?

I modelled my circuit in Falstad and finally got it to work. I'm starting to emerge from the fog a little bit, with the help of you and Merlin. I was amazed to see how easy it is to set the maximum current, independent of voltage (Vds), with one resistor. Transistors are amazing.

 

[andrewRneumann]

It looks like you have your solution. If you decided you need a tube to do this job, check out a Gibson GA-86. It uses a 6V6 as a linear regulator to get 340V screens from a 415V source. The Gibson GA-100 bass amp uses an 0C2 regulator, though I think those have a fuxed voltage drop.

I have a Newcomb Audio KX-30 PA amp on the shelf waiting for some future free time that has an adjustable screen voltage. It also uses a 6V6 regulator. Radiomuseum.org has a schematic. https://www.radiomuseum.org/r/newcomb_kx_30.html

That's cool. I actually have a few 0A2 tubes that I experimented with in shunt regulation. I got very excited when I saw that purple glow. In the end, the current limits of the tube brought me back to Zeners, and then the huge amount of wasted heat brought me to series regulation, and that brought me to MOSFETs. So here I am dabbling in SS like a kid with a marshmallow on a stick.

I would love to find a good excuse to use a gas regulator though. Very cool.

 

[andrewRneumann]

Mouser order is in. I will post photos and results when I get this cobbled together. Thanks again @Ten Over !!

 

[andrewRneumann]

Success!

The MOSFET voltage follower is awesome! I installed a trim pot so that I can set any output voltage I desire. B+ is running about 410V at full drive. I have it set at 56%, so B+2 at full drive is about 230V. The only sag is from the PT... no extra sag from power supply resistors--because there are none! (Unless I want to add some, but now I can choose exactly the amount I want.) I estimate this follower consumes about 0.9W at idle and about 8W at full drive.

It wasn't too hard to build. Everything fit on a 3/4" x 2-1/2" turret board (except for the MOSFET w/ heat sink.) Don't think I need a heat sink... played it pretty hard and heat sink was warm but not very hot at all. Will probably go with chassis mount when permanently installed.

I love it when something works on the first try. Thanks everyone for the input.