Miniature 175V SMPS thread

[PRR]

> If I need 350V ... not a problem to use 2 diodes in voltage doubling configuration?

Won't work. The raw output is NOT symmetrical AC like you get from a wall-transformer. Your other diode has nothing to catch.

How does it work? Get a loooong fishing pole. Clamp the handle to the dock. Pull the rod down 12 inches. Let go. The tip will fly up above level. If you do it right, it can fly-up more than 12 inches, maybe more than 120 inches. This switcher does the same: the transistor flicks the coil with 12V and it flies-up to some higher voltage, which we catch with a diode. For maximum efficiency with real-world parts, you usually only ask for 1:10 fly-up: 120V. Asking for more makes your losses much higher. However the "better way" is more and more-tricky parts. Cliff has accepted "modest" efficiency to get simplicity and 1:15 to 1:20 fly-up.

It's as good as a 1-coil 12V converter gets.

Before you think of it: no, you can't "stack" two of these modules to get 350V, because the output is not isolated. Ah, it would work with TWO 12VDC isolated bricks into two modules, but be careful which brick you get your heater from.

And 12AX7 will usually run very fine with 175V, as long as you don't expect to drive 400V power tubes.



> doesn't it seem funny using a switching power supply ...to reduce 120Vac to 12Vdc and then bump that back up to 175Vdc

You forgot "isolate the audio circuit from wall wiring".

You can turn wall 120VAC into 160VDC VERY simply with a rectifier and cap. However if you touch the audio circuit (guitar jack) and PA mike or concrete or radiator, you may die.

You ALWAYS want a transformer to isolate you from the wall-power.

Classically we wind a big chunk of iron working at 50/60Hz, paying full attention to insulation and spark-over.

If we had 100KHz wall-outlets, the transformer could be far smaller/lighter. But it still needs to be wound with full isolation and insulation. Only that gets tougher in small size.

The problem is very solvable. The 12VDC wall-bricks, also PC power supplies, use a 100KHz transformer for isolation as well as voltage conversion. I've looked at designing my own, and decided the design and the for-sure SAFETY issues were not worth risking my life. There's guys who do this for a living, and other guys who get paid to check their work. I can buy a fully-approved brick for under a buck a Watt. Even with the second voltage conversion, and the general silliness of double-conversion, it still is the best way for normal people to do it.



> Could this smps with its 100ma power a small tube amp like a fender Champ?

"Like" a Champ, sure. But not with 6V6 or EL84 which need over 250V to make full power. Use the AC/DC Radio tubes: 25L6, 50C5, etc. These are built with fat cathodes and will make a whole Watt with 110V on the plate. A few are rated over 150V and as much as 10Wpdiss. Two 25L6 or two 6Y6... heck, one 6Y6 at 200V 75mA 2.6K load will pump 6 Watts which is Champ turf.

> some problem might rise with tube heaters

Since Cliff's plan starts from isolated 12VDC, as armstrom says, you just find 12V-heater tubes. Uh, that's actually a problem: the AC/DC radios ran five tubes in series from 110VAC. In series they all got the same current so the heater voltage was adjusted for the size of the cathode: 35V on the rectifier, 50V on the power tube, 12V on each small tube. And any of these tubes pulls more current and less voltage than Cliff's supply could deliver, even if you were desperate enough to double-convert for a job as simple as heating.

There were a few 6.3V versions, notably 6Y6. But Cliff's supply won't get to 175V at good current with a 6V supply. And the 6Y6's heater current is SO high (1.2A!) that it would be mighty wasteful to drop 12V to 6V. (And a resistor is all you need/want.)

There is 12FX5 (look for 60FX5 data). Typical performance is 110V, 47mA, 3K, 1.3 Watts out. At 150V 36mA (plate) 3.2K load it can make 2W. So two 12FX5 in 1.6K load fed 155V 92mA could be a whopping 4 Watts, "like" a 5W-6W Champ. The pair plus your 12A?7 can be heated from the raw isolated 12VDC, 0.35A needed.

But why? The cost of a good 12VDC couple-Amp brick plus Cliff's module is not much less than a Champ PT. It may be a pound lighter, but you still need an OT and a speaker: it won't be iPod light. The 12FX5 is stunningly cheap, but you'll want to re-diddle stage gains to account for the lower B+ voltage (and the 'FX5's high sensitivity). Quicker and as-cheap to go buy a Valve Jr and cut the nastiness out.

Be realistic. Build the small-tube end of a Champ or Dyna or other small tube circuit. Then a 12VDC brick and an internal voltage-booster makes sense. When you need big tubes and big audio transformers, KISS, use a 60Hz HV PT.

[calico]

Can I power my 3 pieces 6DJ8 Tubes amp ? running 260Volt ?

this tiny SMPS can handle 200mA ?

can not wait the finish board


Cal

[Cliff Schecht]

Sorry this has been on the back burner for so long guys, I've been incredibly busy with school/work plus getting my internship at TI setup and other various crap. I'm going to do another version of this board soon and hopefully it will be the last revision. I think I will increase the size of the board just a bit so that I can switch to through-hole components, SMD boards are a nightmare with no solder masks!

[head_spaz]

BUMP

Has the sun set on this project?

[mensur]

Here is the 320VDC/20mA SMPS which is sufficient for driving 3 x 12AX7's(assuming that maximum current consumption per triode is 3mA, so, 3mA x 6 triodes = 18mA, but normally triodes consume about less than a 1mA). MOSFET has to be heatsinked (cause it will dissipate about 7W of power, and according to datasheet, temp. will rise aboit 60-80C degree for every 1W).

One more thing, I didnt found anything about error amp compensator(i.e. pin 1.), so I just copied it from Cliff's SMPS, maybe he can tell us how we can determine it?

Here are design notes that I found on various app notes:
Vin = 12V
Vout = 320 V
fmin = 100 kHz
Iout = 20 mA

ton/toff = (Vout + VFD1 + VFD2) / (Vin - VsatQ1 - VsatQ2) = 30.89
ton(max) + toff = 1 / fmin = 1 / 100kHz = 0.00001

toff = (ton(max) + toff) / ((ton/toff) + 1) = 0.000000313S
ton = (ton(max) + toff) - toff = 0.000009687S

CT = 4.0 x 10-5 x ton(max) = 0.0000000003 / 387.4pF / 470pF
RT = 1.72/(CT x Fmin) = 1.72 / (470pF x 100kHz) = 36595R / 33K

Ipk(switch) = 2 Iout x ((ton/toff)+1) = 1.2756A

Lmin = ((Vin(min) - VsatQ1 - Vsatq2) / Ipk(switch)) x ton = 0.0000789 = 78.9uH / 125uH

I'pk(switch) = ((Vin(min) - VsatQ1 - Vsatq2) / Lmin) x ton = 0.805A

Rsc = 0.33 / I'pk(switch) = 0.33 / 1.0075 = 0.327 / 0.409R / 0.39R

Vripple(p-p) = ( Vout / Vref) x 1.5x10-3 = (320 / 2.5) x 1.5x10-3 = 0.192V

Co(min) = ( Iout /  Vripple(p-p) ) x ton = (20mA/0.192) x 0.000009687 = 0.000001009 / 1uF I'm going to put 4.7uF.

R2(fb) = R1 x ((Vout/Vref) - 1)) = 1.8k x ((320/2.5) - 1)) = 228K / 220K.

[Cliff Schecht]

Ugh, this project needs to get finished.. It's been on the backburner for too long and I now have the time to actually work on it. The design I first posted (and the boards) have some design issues that I need to resolve though (one of them is very fundamental too ). I will post more later in the week. I may end up having boards for this made as etching your own double sided boards for SMD's is quite a pain.

Mensur: With the compensation, it really depends on the topology. Does your circuit ever go continuous or over 50% duty cycles?

[zambo]

hey guys, I am currently building an amp using a 6k6gt powertube at 200 volts with a 12at7 preamp all powered by a "flyback" smps purchased off evilbay for the low low price of 13 bucks. It seems to work ok an my firefly type projects but this one doesnt seem to want to work right and i am wondering if its a lack of current? I dont understand how to read the charts on duncan amp pages and my measuring devices are primitive at best. Can anyone tell me how many milli amps a 6k6gt will draw at 200v on the plate and screen ( or whatever the other thing is that wants voltage) I know the heater wants .4 amps and I am getting that directly from the 9 volt ps dropped through a 7805...I know its a bit shy for the heater but its the option I had...

I am getting an awsome sound at about 2 watts or so I would guess ( from the amount of dirty looks I get from my family....) the only thing is a persistant high end noise the appears on the lower notes and hangs on for a sec after they stop being played...

I would love to get this thing working and debugged as I could actualt gig in my blues band with it and it fits in a 6x3x2 enclosure and sounds great with an 8 inch jensen. Do you think I need a new power supply? here is what i am currently using     http://cgi.ebay.com/Nixie-power-supply-1363-/140426808956?pt=LH_DefaultDomain_0      any help would be sweet!

[Cliff Schecht]

You could be running into a whole slew of problems. If you are worried about the current, remove the HV connection and place a current probe in series with the HV before it hits any tubes (you measure current in series). Or you could place a 1 Ohm resistor in series with the B+ and measure the voltage drop (in parallel), the voltage drop will be exactly equal to the current if the resistor is at 1 Ohm.

It sounds like you are running into the same stability issues I had with my first few SMPS designs. It's important to know what the switcher is comfortable driving, it could be having issues with the big inductive load that makes up your output transformer/speaker combo. Unfortunately, you need more than some basic tools to do any decisive debugging. A scope to look at the switching waveforms as the noise happens would help out immensely. The problem could be that lower frequency notes take more energy to amplify and are pulling too much current. Another thought is that the power supply probably needs more filtering before it hits the transformer/plate resistors, and maybe an adjustment of the feedback network as well if possible. You could add more capacitance to the feedback to really cut down the feedback loop frequency response which helps to kill any instabilities at the expense of dynamic response (how fast the switcher can react to changes in current/voltage needs).

Post some pictures of the build and perhaps take some basic measurements (power tube current, all your B+ voltages, schematic, etc.). The more you give us, the easier and quicker we can help you.

[zambo]

@ Cliff, Thanks for the rapid response. I will get some pictures up asap. Should be good for a laugh if anything ( i am a total hack bro...). I n the meantime since i dont have any of the right tools I am thinking of putting in a couple of resistors at the plate and screen. Maybe like 100k at the plate and 150k at the screen? ( yes i am totaly guessing ) this should limit the amount of voltage going to the power tube thus going a little easier on the supply making it more stable? Any thoughts on the values or am i just barking up the wrong tree?

[nico.verduin]

With 100K resistors tot the plat and 2mA of current, all the voltage will fall over the resistor leaving none left for the actual tube itself.
You could bias the power tube a lititle colder using a bigger cathode resistor (to pin 8 of the 6V6).

[Michael Allen]

Where did the trim pot get connected? It looks like after D1 and C3, before c9? Just an attenuator?