Author Topic: High Voltage Bench Power Supply Design  (Read 15849 times)

0 Members and 1 Guest are viewing this topic.

Offline H713Topic starter

  • Regular Contributor
  • *
  • Posts: 180
  • Country: us
High Voltage Bench Power Supply Design
« on: January 25, 2019, 04:11:56 am »
One of my hobbies is designing and building tube-based audio equipment. I'm at the point where my Eico 1030 tube-based high voltage power supply isn't really cutting it anymore for a few reasons:

1) At this point it needs a fairly hefty investment in parts and labor (and that means my time), including about $50 worth of tubes and way more time than I want to put into a piece of equipment that doesn't really fit my needs.

2) The 150 mA maximum current is a limitation when prototyping output stages of larger amplifiers. I'd really like to have closer to 300 mA continuous.

3) It's big in an awkward way (tall), and bench real estate is at a premium.


This has led me to the decision to build a new power supply that is better suited to my requirements. In a move that is likely to be controversial, I have decided to go with a solid-state based design using MOSFETs as the series pass element. Unlike many a threads on this, I have no illusions that a 500V 10A MOSFET can handle 10A while dropping 500V across it (I have yet to see a TO-247 device that can dissipate 5 kW). I will likely need multiple devices in parallel in order to get enough current with a comfortable safety margin.

In the interest of simplicity, I have decided to use an LR8 as the voltage reference. This should get me a maximum output of around 450 volts. A few key design specs here:

-It doesn't need to be completely idiot proof. It's a high-voltage power supply, and that in it of itself is anything but idiot proof.

-It should go from 80 to 450 volts (more is better, but the LR8 is limited to 450).

-It should be able to handle 300 mA continuous.

-It should be very quiet (electrically, that is. There will be a fan).

-It should be current limited. If a device under test were to short, It should current limit it so that it doesn't immediately blow the pass elements. In this situation it would dump the full 500V output from the rectification stage into the shorted load- not a good situation, creating lots of smoke and sparks. The filter capacitors will not be very small, so blowing the MOSFETs to pieces would not be hard without some form of protection.

The power transformer in question is a Tektronix 120-037. Information can be found on the web for those who are interested, however for me to get about 500V rectified output from it at sufficient current is easy, and I already have the filter capacitors and rectifiers on hand. This will likely be a cheap project, as aside from the FETs and the LR8, I should have most of what I need in my "junk box".

Looking at the DC SOA, I think that a pair of these FQA8N90C-F109 should be sufficient for the series pass element with a reasonable safety margin.  https://www.digikey.com/product-detail/en/on-semiconductor/FQA8N90C-F109/FQA8N90C-F109-ND/1954558

I know that at a lower voltage output (say 100V), this will be dissipating a lot of heat. I'm planning to mount the FETs to the heatsink with a clamping bar. I also plan to take the time to ensure the heatsink is actually flat (I will put it on the mill and skim-cut it with a fly cutter if need be). In order to maximize heat transfer, there will not be any mica spacers. This means extra care will have to be taken in chassis design since the heatsink will be hot (both electrically and physically).

That pretty much leaves my current limiting circuit, which I'm not quite sure how I should go about. It needn't be adjustable, I was planning to leave it fixed at around 500 mA.  I have a distinct feeling that an LM317 is not going to be able to take the voltage drop necessary. Most of the designs I've seen for high-voltage current limiting circuits use a BJT, but I'm concerned that I will run into issues with secondary breakdown. If I could use the same MOSFETs as for the pass element, that would be great, but I haven't seen a whole lot of designs that do this.

I would like to keep this reasonably simple. I do not plan to design a PCB for this. Perf-board and point-to-point is the plan.

I'm sure I'm crazy, but this is so far what I've come up with. I have thought very hard about using tubes for the pass element- there are certainly advantages (usually they don't fail short like a FET) but at the desired output it would simply take up too much space. I'd love to hear suggestions on how I should proceed. Hopefully I'll have time to draw up a schematic with my idea this weekend.

Sorry for the long post.
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9320
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: High Voltage Bench Power Supply Design
« Reply #1 on: January 25, 2019, 05:24:17 am »
What about use a variac to vary the voltage?
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 7678
  • Country: ca
Re: High Voltage Bench Power Supply Design
« Reply #2 on: January 25, 2019, 05:45:52 am »
A few threads on HV power supply design have crash and burned here. Good contributors get into a flame war, edited/deleted posts, banned etc. I may have soot on my feathers.

https://www.eevblog.com/forum/projects/looking-for-yours-opinions!-hv-stabilized-power-supply/
https://www.eevblog.com/forum/projects/200v-200300ma-power-supply/
https://www.eevblog.com/forum/beginners/a-good-homebrew-500v-1a-power-supply/
https://www.eevblog.com/forum/projects/planning-to-build-500v-tube-psu-help-needed/
https://www.eevblog.com/forum/beginners/transistor-vs-mosfet-for-linear-power-supply/

Vacuum tubes are economical, robust and much simpler to design-in for pass elements compared to solid-state. This digresses into a religious war of sorts. Cheap, easy and antique is a couple pentodes.

Modern/elegant is all solid-state but unfortunately "trial by fire" so design work is tedious. Very few MOSFETS are rated for power/decent SOA in the linear region. You'll have to series or parallel multiple parts or forgo a completely linear design and make a HV SMPS or a hybrid with a pre-regulator or transformer tap switching/variac. Or look for an off-the-shelf old BK Precision 9185 series, or Xantrek (Sorensen/Ametek) or Kikusui etc.

It all depends on your budget, patience and expectations. A HV linear supply is never going to be compact. This is not a simple or cheap project.
In old threads, requirements were not nailed down and a few hundred volts or mA here or there can add 10x the cost/complexity.

There's no shortage of theoretical designs and optimistic Spice simulations. I did not see a design get carried to fruition.
 
The following users thanked this post: 001

Offline coromonadalix

  • Super Contributor
  • ***
  • Posts: 7001
  • Country: ca
Re: High Voltage Bench Power Supply Design
« Reply #3 on: January 25, 2019, 08:08:07 am »
Some people reported success with electrophoresis  psu's  ??
 

Online Gyro

  • Super Contributor
  • ***
  • Posts: 10172
  • Country: gb
Re: High Voltage Bench Power Supply Design
« Reply #4 on: January 25, 2019, 09:19:06 am »
Some people reported success with electrophoresis  psu's  ??

IF you can find one that isn't center grounded, without ground leakage, no load sensing etc. You probably need to do something about their voltage regulation response too (mine is still waiting for that experiment!):

https://www.eevblog.com/forum/projects/300v-dc-supply/msg1926577/#msg1926577
Best Regards, Chris
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1826
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: High Voltage Bench Power Supply Design
« Reply #5 on: January 25, 2019, 09:30:57 am »
Hi,

some comments:

- The FET you have chosen is not made for linear operation. You need an IXYS linear MOSFET made for this type of load, and
probably you need to run some in parallel. IXTH2N150L would be a choice.

- current limiting is fine, but in case of a short circuit you will dissipate ca. 200W. Maybe some preregulator is a smart idea,
 or transformer tap switching if this is possible with your transformer.

- I have made a 500V supply with 30mA. You could change my circuit to 300mA by replacing the pass MOSFETs
Link here:
https://electronicprojectsforfun.wordpress.com/power-supplies/high-voltage-lab-power-supplies/a-500v-30ma-variable-linear-power-supply/

- Play safe!

Wolfgang
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20357
  • Country: gb
  • 0999
Re: High Voltage Bench Power Supply Design
« Reply #6 on: January 25, 2019, 09:52:54 am »
What's wrong with the FQA8N90C-F109? It's clearly specified for DC linear operation. Refer to figure 9 on page 4 of the data sheet.
https://www.onsemi.com/pub/Collateral/FQA8N90C_F109-D.PDF
 

Offline beanflying

  • Super Contributor
  • ***
  • Posts: 7375
  • Country: au
  • Toys so very many Toys.
Re: High Voltage Bench Power Supply Design
« Reply #7 on: January 25, 2019, 10:12:14 am »
There is a  few of these on evilbay in the USA. Very unlikely you could buy or even scrounge the bits for this sort of price. eBay auction: #173757684650

Coffee, Food, R/C and electronics nerd in no particular order. Also CNC wannabe, 3D printer and Laser Cutter Junkie and just don't mention my TEA addiction....
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1826
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: High Voltage Bench Power Supply Design
« Reply #8 on: January 25, 2019, 10:18:40 am »
What's wrong with the FQA8N90C-F109? It's clearly specified for DC linear operation. Refer to figure 9 on page 4 of the data sheet.
https://www.onsemi.com/pub/Collateral/FQA8N90C_F109-D.PDF

You're right, I have seen the DC in the SOAR, only I have no trust in parts developped for switching purposes. Fairchild parts also had a DC SOAR, and they blew up nevertheless. In a linear HV PSU, the pass transistor is the most critical part of the circuit. If this shorts out, the full high voltage will appear at the DUT, plus the charge stored in the smoothing cap after the rectifier. This can be many Joules of energy, normally killing your DUT (and hopefully not you too) instantly. I would play extremely safe here.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: High Voltage Bench Power Supply Design
« Reply #9 on: January 25, 2019, 11:32:15 am »
I have a few FQA9N90Cs, I'll test DC SOA sooner or later.

Good choice, I found they had the least cost per watt of any transistor with DC SOA and the ratings needed for a project.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1826
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: High Voltage Bench Power Supply Design
« Reply #10 on: January 25, 2019, 11:47:19 am »
... testing the SOAR is a good idea. A test would run a normal load, plus then pulses of the current that would occur when the output is shorted. A very robust load is recommended, just in case the test fails.  >:D
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20357
  • Country: gb
  • 0999
Re: High Voltage Bench Power Supply Design
« Reply #11 on: January 25, 2019, 12:39:23 pm »
What's wrong with the FQA8N90C-F109? It's clearly specified for DC linear operation. Refer to figure 9 on page 4 of the data sheet.
https://www.onsemi.com/pub/Collateral/FQA8N90C_F109-D.PDF

You're right, I have seen the DC in the SOAR, only I have no trust in parts developped for switching purposes. Fairchild parts also had a DC SOAR, and they blew up nevertheless. In a linear HV PSU, the pass transistor is the most critical part of the circuit. If this shorts out, the full high voltage will appear at the DUT, plus the charge stored in the smoothing cap after the rectifier. This can be many Joules of energy, normally killing your DUT (and hopefully not you too) instantly. I would play extremely safe here.
As I said in one of those other threads: never rely on semiconductor or thermionic devices to protect against electric shock. Always assume the output of your power supply has the potential to be at a hazardous voltage, even when it's set to <50V.

It should have overvoltage protection, but even so, it shouldn't be relied on to protect the user from shock, only the DUT.
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1826
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: High Voltage Bench Power Supply Design
« Reply #12 on: January 25, 2019, 12:44:56 pm »
Even if the PSU works just fine, its output voltages can be lethal, as with any HV tool. Use caution even when you set the output at low voltages. Reasons see above.

Old HV slogan:

Your first error is your last error.  >:D
 

Offline H713Topic starter

  • Regular Contributor
  • *
  • Posts: 180
  • Country: us
Re: High Voltage Bench Power Supply Design
« Reply #13 on: January 25, 2019, 06:16:12 pm »
Perhaps the key aspect here is the fact that I have most of the parts on hand at this time. All of the costly parts (Transformer, filter capacitors, panel meters, chassis, connectors, etc) is stuff I have on hand. Yes, I am aware that this FET is designed for switching. That is why I'm planning to put at least two in parallel, if not more, which should keep it well within the DC SOA. The idea is that each device will be run conservatively enough that even though it's not really designed for linear operation, it should be fine. A circuit breaker or HV fuse on the output will be implemented as well to stop anything too stupid from happening.

The current limit is probably the hardest, but I'll spend some time digging around for a reasonably robust way of doing it.

I pretty much assume (from a safety standpoint, that is) that if the output is capable of producing 500V, then I'm going to treat it as if it always does.

Also, the key aspect is that I recognize the fact that the pass elements could short. That will always be in the back of my mind while I'm using this power supply.
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1826
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: High Voltage Bench Power Supply Design
« Reply #14 on: January 25, 2019, 06:20:05 pm »
What you could do for a safety measure is to implement an adjustable overvoltage crowbar at the output.
 

Offline H713Topic starter

  • Regular Contributor
  • *
  • Posts: 180
  • Country: us
Re: High Voltage Bench Power Supply Design
« Reply #15 on: January 25, 2019, 08:30:12 pm »
I was thinking about doing a crowbar, however I'm wondering if you had anything in mind in terms of implementation.

The first idea that comes to mind is to use a two-gang pot for the voltage adjust, the second of which forms an adjustable voltage divider. That along with a zener could be used with a beefy triac if the output voltage goes noticeably higher than it's supposed to. What I'm not a huge fan of about this is that it's putting a fair bit of trust in the wiper of a potentiometer.

The second idea is a fixed crowbar circuit for say 500V. If the adjustment is limited from 0-450V, and it goes up to say 500 (indicating a pass element short), then the crowbar would be triggered (again using a zener, but this time with a fixed voltage divider).

Attached is the world's ugliest schematic to try and illustrate my idea.
Is this what you had in mind?
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1826
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: High Voltage Bench Power Supply Design
« Reply #16 on: January 25, 2019, 08:37:04 pm »
You got the idea.

In the circuit below:

https://electronicprojectsforfun.wordpress.com/power-supplies/high-voltage-lab-power-supplies/a-1kv-50ma-linear-power-supply/

I have created an electronic "fuse" that switches off a preregulator output in case of *overcurrent*.
You could do something similar for *overvoltage".
 

Offline 001

  • Super Contributor
  • ***
  • Posts: 1170
  • Country: aq
Re: High Voltage Bench Power Supply Design
« Reply #17 on: January 25, 2019, 08:38:37 pm »
oh, man

it is long tale
no easy way to make it with transistors
check Heathkit units at ebay or rebuild Your own clone
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1826
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: High Voltage Bench Power Supply Design
« Reply #18 on: January 25, 2019, 09:06:26 pm »
Correct. Reliable transistor (or MOSFET) high voltage supplies are not as simple as they look.
But its an interesting subject, IMHO.
 

Offline H713Topic starter

  • Regular Contributor
  • *
  • Posts: 180
  • Country: us
Re: High Voltage Bench Power Supply Design
« Reply #19 on: January 26, 2019, 01:39:26 am »
So I've drawn up a schematic with my idea as it stands now. I may choose to parallel a third transistor, though I do suspect that two would be more than enough. I still haven't worked out my current limiting circuit, so if anyone has an idea as to how to go about it I'm all ears.

One thing I am planning on implementing (though I didn't draw it in the schematic) is to have a "50-250" and a "250-450" voltage switch. In the 50-250 mode, it shorts across R1 and R5. This does several things. Firstly, it means I don't need to use a 500K pot for the voltage adjust (more precise setting), and it will limit the amount of power dissipated by the pass elements. This combined with a current limit (if I can come up with something suitable) and the crowbar circuit makes me feel much better about the reliability of the FETs.




Edit: Do NOT try to build this schematic as shown- Testing at a later date revealed that the FETs cannot meet their published specs and a pair of them in parallel will not be able to handle the load. More viable schematics are shown towards the end of page 3.
« Last Edit: July 14, 2019, 04:21:46 am by H713 »
 

Offline Wolfgang

  • Super Contributor
  • ***
  • Posts: 1826
  • Country: de
  • Its great if it finally works !
    • Electronic Projects for Fun
Re: High Voltage Bench Power Supply Design
« Reply #20 on: January 26, 2019, 02:27:45 am »
The crowbar is OK, maybe it needs a small cap across the resistor so it does not fire from just a small spike.
Manuall switching transformer ranges is possible, but unelegant. You could do this electronically.

Look here:

https://electronicprojectsforfun.wordpress.com/power-supplies/preregulator-circuits-for-linear-power-supplies/

for some ideas.
 

Offline H713Topic starter

  • Regular Contributor
  • *
  • Posts: 180
  • Country: us
Re: High Voltage Bench Power Supply Design
« Reply #21 on: January 26, 2019, 03:08:57 am »
Thanks, I didn't even think of switching transformer taps. If I use the Tek power transformer which has a lot of windings, that could work quite well. The voltage doubler on the 2.5 ohm 155V winding (which would be put in series with two of the seven 6.3V taps this thing has) was just the easiest way to get around 500 volts- but this is definitely a better way to do it.

Electronic switching would be great, but it wouldn't solve the potentiometer issue. I could, of course, just use a fine and coarse adjustment system, which would be pretty easy to implement.

Regardless of whether I go for the electronic switching or doing it manually, I do like the idea since it reduces the stress on the FETs.

One thing that should be considered with this design is that if the wiper on the potentiometer were to go open, the power supply would jump to it's highest output setting. It wouldn't be high enough to trigger the crowbar circuit, however, because it would still be within the power supply's acceptable output range. This is a problem I've seen on most of these designs (including many tube ones). Potentiometer failures aren't super common, but they do happen and I would hazard a guess that it's more likely than having the FETs blow without triggering the crowbar. The dangers of this could be mitigated however by having the crowbar circuit change whenever the transformer taps are switched. While an open pot wiper would still be a bad situation, at least the power supply wouldn't dump 500 volts into a load intended for 100.

A current limit fixed at 750 mA would help mitigate the damage done if something were to fail. 750mA is still way more than enough to kill you, but a short probably wouldn't be quite as spectacular.

Edit: Forgot to attach schematic.
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
Re: High Voltage Bench Power Supply Design
« Reply #22 on: January 27, 2019, 12:46:26 am »
Background: I have a Kepco ABC-425 power supply (400 V @ 50 mA) that has the difficult to get 8068 regulator tube.  It was built in 1973 and except for the tube is all solid state.  I've been toying with the idea of replacing the tube with FETs should it expire.  I'd sort of known that switching FETs weren't the best choice for linear applications even though in the late 70s and 80s the SOA graphs said they should work well.  Apparently, the problem is Electro-Thermal Instability.  I've attached some files showing this a bit better.  NASA plus the FET manufacturers have more info on this phenomenon.

To get back to the OP's design:  When I look at the attached graph, I see that ETI is more likely at low current and high voltage.  This tells me that it's probably better to put the pass FETs in series rather than in parallel.  It's not quite as easy as paralleling them but it can be done with a voltage divider.

As others have said, a short circuit is bad news.  This can happen if a capacitor fails.  When I look at the latest schematic, I don't see any current limiting.  If the output is shorted or the crowbar is fired, C6 will discharge through the 15 V zener and will probably short it out. Depending on the zener's series resistance, it may momentarily increase the drive to the FETs, causing them to dump the DC bus into the load as well.

I work with Variable Frequency Drives using IGBT modules that incorporate cycle by cycle current limiting, quite often by monitoring the IGBT's source current.  Most of these drives can handle a direct short to ground without damage.  Applying this to the OP's design, I would add a fast current limiter that would monitor the voltage across the FET source resistors, and reduce gate drive if too high.  This could be done with an NPN transistor and a few resistors.  If the pass FETs were put in series as I suggest above, I would put the current limiter on the upper FET and have it latch off when there was gross overcurrent.  This could be done with a PNP-NPN latching circuit.  A more precise variable current limit could be implemented around the lower FET if desired.

For what it's worth,
« Last Edit: January 27, 2019, 05:00:24 pm by duak »
 

Offline 001

  • Super Contributor
  • ***
  • Posts: 1170
  • Country: aq
Re: High Voltage Bench Power Supply Design
« Reply #23 on: January 27, 2019, 09:09:22 am »
I have a Kepco ABC-425 power supply (400 V @ 50 mA) that has the difficult to get 8068 regulator tube.  It was built in 1973 and except for the tube is all solid state.  I've been toying with the idea of replacing the tube with FETs should it expire.  I'd sort of known that switching FETs weren't the best choice for linear applications even though in the late 70s and 80s the SOA graphs said they should work well.  Apparently, the problem is Electro-Thermal Instability.  I've attached some files showing this a bit better.  NASA plus the FET manufacturers have more info on this phenomenon.

When I look at the graph, I see that ETI is more likely at low current and high voltage.  This tells me that it's probably better to put the pass FETs in series rather than in parallel.  It's not quite as easy as paralleling them but it can be done with a voltage divider.

As others have said, a short circuit is bad news.  This can happen if a capacitor fails.  When I look at the latest schematic, I don't see any current limiting.  If the output is shorted or the crowbar is fired, C6 will discharge through the 15 V zener and will probably short it out. Depending on the zener's series resistance, it may momentarily increase the drive to the FETs, causing them to dump the DC bus into the load as well.

I work with Variable Frequency Drives using IGBT modules that incorporate cycle by cycle current limiting, quite often by monitoring the IGBT's source current.  Most of these drives can handle a direct short to ground without damage.  I would add a fast current limiter that would monitor the voltage across the FET source resistors, and reduce gate drive.  This could be done with an NPN transistor and a few resistors.  If the pass FETs were put in series as I suggest above, I would put the current limiter on the upper FET and have it latch off when there was gross overcurrent.  This could be done with a PNP-NPN latching circuit.  A more precise variable current limit could be implemented around the lower FET if desired.

For what it's worth,

I need  Kepco schematic to help You
I googled it but no results
Can You post Yours circuit diagram here?
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
Re: High Voltage Bench Power Supply Design
« Reply #24 on: January 27, 2019, 04:53:58 pm »
001, I don't have a schematic for the Kepco supply.  It is working fine so I don't need it.  If the vacuum tube fails the circuit is simple enough to trace out.  I mentioned it because I have tubes of IRFP360 MOSFETs (rated for 400 V) and I wondered if I could use them for something.  I looked into using them for regulators and loads and found the research on Electro-thermal Instability.  Thank you for your offer which made be realize that my post was confusing.  I've edited it to hopefully make it clearer.

As an aside, I was thinking about the size of a vacuum tube versus a transistor.  The active volume of the tube is a few cc and the plate area is a few square cm.  The transistor die is a few square mm in area with the actual active volume just the top tens of microns.

Cheers,
« Last Edit: January 27, 2019, 05:15:25 pm by duak »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf