Miniature 175V SMPS thread

[Cliff Schecht]

I've finally gotten the 175V supply to a place where it deserves it's own thread. I ended up designing the supply for 100 mA of current at 175V (17.5W) so that people wouldn't ever feel like they are pushing the supply towards its limit (also keeps heat down at "max" load). The max recommended load will be below the actual supply goal (probably 60 mA) so that this unit can be built inside of a case without worry of excessive heat building up inside an enclosure. The design is based around (currently) the UCC3813-2 IC but I'm providing provisions to use similar IC's with very minimal parts changes. It's designed to operate from a 9-15V DC source. Here's the schematic:


Parts List (including resistor/cap packages):
R1: 15k 0805
R2, R7: 0.1 Ohm 2W 2512
R3: 200k 0805
R4: 2.9k 0805
R5: 100k 0805
R6: 100k 1206
R8: Not needed
R9: See notes below
R10: See notes below

C1, C2: 470 pF 0805
C3: 22uF 200V low ESR
C4: See notes below
C5, C6: 1000pF C0G 0805
C7: .1uF 0805
C8: 10uF Ceramic 1206
C9: .022uF 0805 200V
C10: See notes below

U1: UCC3813-2 (preferred) or UC3843
Q1: Low RdsOn N-Channel MOSFET, I like the ST Micro STP9NM50N as a candidate
D1: UF4004, 4005, 4006 or 4007
L1: 15 uH, 3A + inductor --> Bourns SDR1806-150ML looks like a good candidate and fit for the board
In and Out connector: 158-P022ESDVP-E Eagle connector or similar screw block terminal

Here's a picture of the layout as it currently is:


At 2" x 1", the board is really made to fit inside a 1590 style case without much fuss. All of the parts mount directly to the top of the board and there are (currently) two jumpers. The one on the bottom is put in place to give the IC and associated components a stronger ground plane and I'm debating on the need of it. I may take it out depending on how my first prototype build goes. The FET is a TO-220 package that will need heatsinking.

With the IC I tried to go with the more modern UCC3813 but if this isn't available, a UC3843 can be used as well. The only differences is that the UCC3813 has an internal 12V Zener diode shunt regulator that limits the top VCC (hence the 1/2 W 100 Ohm resistor in series with the input VCC to the IC) whereas the UC3843 can get up to a 30V VCC without issues (meaning short R10). Also, the UC3843 doesn't have current blanking on the current sense pin and will require the R9, C10 parts to be placed (small RC filter of 10 Ohms and 1000 pF). If using the 3813, R9 gets shorted and C10 isn't stuffed. Also I've found that with my prototype build, a snubber circuit isn't necessary. I've still placed parts (R8, C4) just in case one is needed, but if not then these parts don't need to be stuffed.

I'm hoping to have this board verified by next week and when I do, I will post a parts list that includes Mouser part #'s and total cost (should be under $15). We're also talking about stocking these boards at PAiA if there is enough interest, but if we do this then I will probably try to shrink the layout more and use a double-sided board.

Thoughts/comments/criticisms?

[g.]

Great news !

is there any chance to have a board design with non surface mount components ?
60mA is large enough !

[frequencycentral]

Hey Cliff, good to see this project 'lift off'. I'm going to have to get a parts order together before I try it. Interesting that you chose an SMD layout - that puts it out of reach for me (and maybe some others too) but I'll work up a perf/PCB layout for through hole components based on your layout. Is this going into your 6BM8 based build?

[R.G.]

Nice design, Cliff.

Quite apart from that, have you considered the safety issues with even a properly-operating board like this?

[petemoore]

 Thoughts/comments/criticisms?
 Looks pretty tight ! As in small footprint...
 What is the ~cost of building a PS like this ?

[punkin]

HMmm...I'm seeing a lot of possibilities here.


THANKS!

[Ice-9]

I will have to give this one a try Cliff. very good looking layout.

[Cliff Schecht]

Nice design, Cliff.

Quite apart from that, have you considered the safety issues with even a properly-operating board like this?

Funny you mention this. When I first finished the supply, I placed my hand across the load resistor to feel if it was getting hot and got a small tingle (175V w/ 50 mA limit). After I went through and redesigned everything for 100 mA, I again placed my finger on the minimum load resistor and got the shit shocked out of me! No joke about the safety issues of a supply like this! I'll definitely attach an adequate warning label (and disclaimer) to the build-ready documents.

Also, I understand you guys not liking the SMD components. I'll try to do a second PCB version with 1/4W resistors and through-hole caps to see what I can get. Size will go up of course, but for some this isn't the issue. I have a large collection of 0805 resistors and caps and so I tend to go for those parts when I do layouts now. They do make DIP versions of the UC3843 and UCC3813 and so going through-hole isn't such a big deal.

Cost should be about $12 with Mouser although using a better quality FET will increase the cost by a few dollars.

Oh and thanks for the kind words guys! I've taken enough away from this community, it's time to give back .

[R.G.]

Funny you mention this. When I first finished the supply, I placed my hand across the load resistor to feel if it was getting hot and got a small tingle (175V w/ 50 mA limit). After I went through and redesigned everything for 100 mA, I again placed my finger on the minimum load resistor and got the sh*t shocked out of me! No joke about the safety issues of a supply like this! I'll definitely attach an adequate warning label (and disclaimer) to the build-ready documents.

Good idea. Just from some safety stuff I was forced through like beef through a hamburger grinder:

20ma DC through the chest is sufficient to clamp the heart down to inaction
20ma of 50-60 cycles starts ventricular fibrillation very efficiently
Skin resistance for dry skin is about 20K
Safety labs declare any accessible voltage of over 42V as hazardous on the face of it
"Accessible" means you can touch non-grounded (to a third-wire safety ground) metal

You may want to make people agree to a legal release before giving it to them. It's a PITA, but less than the PITA of even one legal action brought by the non-technical survivors of a less-than-clever builder who does something dumb. Anytime lawyers touch anything it gets complicated, expensive, and uncertain.   

[Cliff Schecht]

Yeah I'm not sure about where we would stand legally if PAiA sold these boards built, as a kit or as a blank PCB. That's not my department though, so I'll see what the elders say.

It's scary how easy it can be to kill yourself with electricity. Every shock I receive is humbling for sure. I'm shooting a safety video for IEEE next week and will make sure to include some HV warnings.

One of my lab partners once told our advisor "I'm thinking about going into patent law, maybe becoming a lawyer". I'm pretty sure he got a B.

[Ice-9]

Nice design, Cliff.

Quite apart from that, have you considered the safety issues with even a properly-operating board like this?

Funny you mention this. When I first finished the supply, I placed my hand across the load resistor to feel if it was getting hot and got a small tingle (175V w/ 50 mA limit). After I went through and redesigned everything for 100 mA, I again placed my finger on the minimum load resistor and got the sh*t shocked out of me! No joke about the safety issues of a supply like this! I'll definitely attach an adequate warning label (and disclaimer) to the build-ready documents.

I'm not too familiar with this chip but most the PWM chips have a connection so you can fit a switch , thus disabline the HV. One of those little tactile switches with a spring on so that when the effect pedal cover is removed a least removes the hazard of a shock when the cover is removed,, although it doesn't help while building and testing the supply.

[Cliff Schecht]

Working with the HV supply is about as easy as making sure that the power is removed before you do any work. I also like to have a meter watching the HV output at all times so that I have visual confirmation that the PSU is off. Better safe than shocked.

[aziltz]

Working with the HV supply is about as easy as making sure that the power is removed before you do any work. I also like to have a meter watching the HV output at all times so that I have visual confirmation that the PSU is off. Better safe than shocked.

I'm working with 1KV and/or 150A in my lab lately.  Be Careful.

[R.G.]

Working with the HV supply is about as easy as making sure that the power is removed before you do any work. I also like to have a meter watching the HV output at all times so that I have visual confirmation that the PSU is off. Better safe than shocked.

If you've ever done any flying, you've heard the aphorism that there are old pilots and bold pilots, but no old AND bold pilots. 8-)

Darwin may not have gotten every little detail correct, but there is certainly a filtering process that happens when doing things with some risk involved.  Quite a number of the techs that I know who have worked on tube amps for years will keep a clip lead handy and clip one end to the chassis and the other end to the first filter cap + lead. Then they wait a few minutes, and they leave the clip lead there while they are working on the amp.

The very adventuresome ones will make up a clip lead with a resistor, maybe 10K to 50K at several watts to take it easier on the filter caps if they happen to be charged when the clip lead is connected, and to smoke (instead of frying the power transformer) if they forget to remove it before powering up.

It's almost like Mother Nature doesn't care how brave you are, isn't it?   

[petemoore]

  I put 'smoke tape' [electrical tape] around the drain resistor.
  The wire gets a loop around the power switch, then clipped to the chassis.
  That, and the AC input are the two big ones, I like to measure low or no voltage across the big caps for good measure.

[g.]

how about a bleeding resistor ? about 47K from HT to ground for discharging the caps after power off ?
i add one in each tube gear i work on.

[Cliff Schecht]

Every (well-designed) SMPS will have a load resistor on the output to draw a small amount of current. Switching supplies need this minimum load resistor for stability purposes. The minimum current draw is one of design parameters you must know when choosing the feedback loop components.

[Cliff Schecht]

So I went ahead and did a through-hole version of this board. It's a bit longer at 1" x 2.55" but it should still fit nicely in a small enclosure. I had to make more room for the FET to be heatsinked properly compared to the board above where the FET can be soldered in off the board a bit and heatsinked easily. The switch to bottom-side copper was obvious for the through-hole version of this board. Standing resistors really help save on size although I'm not sure how well they'll play out with higher frequencies. Only one jumper on this board and no component value changes (aside from eliminating R7). I'll post both this layout and the SMD version in a printer-friendly version, along with a more detailed parts list, when I get the layouts etched and verified.

This isn't the final version, but a sneak preview nonetheless:

[g.]

Great stuff Cliff !

[MetalGuy]

I would be nice to see how the output looks on the scope. Also to perform some A/B tests with actual equipment.
I would like also to remind you about the discussion concerning the dead batteries, compression etc. I suppose a 4k7-10k resistor after first cap (or maybe didode?) will do the job.
I also have one question. If I need 350V for a pair of 12AX7s for example I guess it's not a problem to use 2 diodes in voltage doubling configuration?