The MAX1771 SMPS Development Thread

[frequencycentral]

I've got some of these MAX1771 to try, I know Mick (Ice-9) has already tried these and I've PM'ed with him to pick his brains, but I thought the topic deserved it's own thread, so here it is.

MAX1771 data sheet: http://datasheets.maxim-ic.com/en/ds/MAX1771.pdf

MAX1771 SMPS schematic: http://desmith.net/NMdS/Electronics/NixiePSU.html

I've redrawn the schematic, partly to go over the whole thing in my head, and partly to rationalise the schematic. I'm pretty sure it's correct, if someone would care to check it against the original that would be great.



I just breadboarded it without a result.    First mistake was that I misunderstood the value of RSENSE, which I read as 50R, it's actually 0.050R - that's tiny, and way lower than anything I have in stock. For the ultra fast recovery diode I'm using UF4004, for the MOSFET I've got IRF740 and IRF840 on hand. I've got a nice selection of inductors. I wonder where I'm going wrong (apart from the value of RSENSE). Should the diodes and MOSFET I have work with this? I also am at a loss to calculate the wattage needed for RSENSE - I need to relearn Ohm's law I guess.

So I've started this thread just to share the information publically, rather than just PM back and forth with Mick. Hopefully it'll also be useful to some other people too. Feel free to chip in anytime!

[R.G.]

For high frequency switching power supplies, even more than for analog audio, *everything* matters. You may have some silly, trivial mistake, just like you or I could make on any circuit. Or it could be that the parasitic inductance and capacitance of the wire to the gate of the MOSFET is letting it oscillate at several hundred megahertz.

Have you read the datasheets on the control chip and MOSFET enough to have them substantially memorized?

[Ice-9]

Hi Rick ,

I see you got started on the Max1771 HV supply. The schematic checks out the same as the one i was starting from except there are 2 missing caps  from pin 2 to ground (100n and 10uf) I think these need to be in your circuit.

The diode you are using will work just fine, infact its the same as i used UF4004 and the MOSFET is also the same IRF740.

Rsense is very important in getting the SMPS to work and you might want to double the resistance from .05 ohm to 0.1 ohm, this will halve the maximum current available. Rsense should be a 1watt resistor.

The 1M5 resistor must be a high voltage type as it will have nearly the full output voltage across it.

I hope this is some relevent info.

[frequencycentral]

For high frequency switching power supplies, even more than for analog audio, *everything* matters. You may have some silly, trivial mistake, just like you or I could make on any circuit. Or it could be that the parasitic inductance and capacitance of the wire to the gate of the MOSFET is letting it oscillate at several hundred megahertz.

I've tried to keep jumping leads to a minimum, and been careful to not cross jumper close to each other. I'm aware that breadboards are not teh best places for SMPS's to live, though the 555 based SMPS I worked on last worked just fine on breadboard.

Have you read the datasheets on the control chip and MOSFET enough to have them substantially memorized?

The schematic checks out the same as the one i was starting from except there are 2 missing caps  from pin 2 to ground (100n and 10uf) I think these need to be in your circuit.

I figured this is just filtering and ripple control, in parallel with the 100uf. My PS is pretty well filtered, but I'll add them to my schemo (and breadboard) when I get time though.

The diode you are using will work just fine, infact its the same as i used UF4004 and the MOSFET is also the same IRF740.

Thanks for confirming that Mick.

Rsense is very important in getting the SMPS to work and you might want to double the resistance from .05 ohm to 0.1 ohm, this will halve the maximum current available. Rsense should be a 1watt resistor.

Yeah - subbing in a 47R or 100R just isn't going to cut it is it?

[Cliff Schecht]

I have no experience with that part so I'm not sure about this, but it looks like there is no compensation on the feedback network. Actually, I'm sure there is no compensation and this is risky business. If you're not careful, you'll accidentally create a fairly high frequency oscillator and/or increase output noise. Typically an extra pole is added to the feedback (in the form of a low-pass filter) to prevent excessive phase shift and potential instability. I would go with what the datasheet says and run the 1771 in bootstrap mode for the best performance. It doesn't cost much more (a few more passives) and makes for better regulation.

There are also some other little tricks that you can do to dramatically decrease the output noise. A snubber circuit across your MOSFET (series RC circuit from drain to source) is used to counteract the high frequency ringing that occurs from the leakage inductance of the inductor combined with the capacitance of the FET's drain. It'll probably be up in the 10-15 MHz range and while it's not necessarily audible, it does decrease the efficiency of the switcher by more than one would think. Another trick you can use is to have a Pi-network on the output. This looks like a cap to ground feeding a power choke feeding another cap to ground. The idea is that at DC frequencies, your caps and inductors look like open and short circuits, respectively. As you start getting higher in frequency, the caps and inductors start looking more and more reactive and shunt switching noise nicely.

I just got the parts in to start on my DIY switcher. Hopefully later in the week or next weekend I'll find time to play around with what I have. The IC used here (MAX1771) is too expensive IMO, I chose a part that was about $0.70 in single quantity from Mouser and they had thousands in stock. As much as I like Maxim's stuff, I usually avoid using their parts in designs that are meant to be inexpensive and/or built by many. The parts can be hard to find and when you do, you pay out the wazoo. Same goes for AD, they make fantastic products but they come at a premium price.

[davidallancole]

Hi Cliff,

I was wondering which chip you are using for your switcher?

[Cliff Schecht]

The UC3842: http://www.st.com/stonline/books/pdf/docs/4299.pdf

[Ice-9]

The UC3842: http://www.st.com/stonline/books/pdf/docs/4299.pdf

I had originally considered using the UC3843 chip which is the same as used in the blackstar valve pedals, but as i had a max1771 already i tried with that. I think one of the advantages after reading the datasheets is that with the UC3843 the switching frequency can be set but with the max1771 it seems the swithcing frequency is controlled by the feedback to the chip itself and is constantly variable.

[davidallancole]

Nice looking chip Cliff.  I can't wait to see your supply.

[R.G.]

One thing I always intended to finish writing up but never got to was a power isolator. 9V in, 9V isolated out.

I've had several one-chip versions of this running, including one that was capacitively isolated, no inductive parts needed. That was fun. 

I think there's some commercial company that finally tumbled to this, supplying a "virtual battery" which is the same thing. It's just an isolator.

Easy enough to do once you think of it. Tricky to get it to work (a) reliably  and (b) quietly.  But a good candidate for postage-stamp PCBs.

[Gus]

http://www.interfacebus.com/Copper_Wire_AWG_SIze.html

Note the ohms per 1000ft.  Take the wire size resistance divide by 1000.  Lets take 38 gauge.  673 ohms per 1000 ft.  673/1000=.673 ohms per foot.  or about .056 ohms an inch

[JKowalski]

One thing I always intended to finish writing up but never got to was a power isolator. 9V in, 9V isolated out.

I've had several one-chip versions of this running, including one that was capacitively isolated, no inductive parts needed. That was fun.  

I think there's some commercial company that finally tumbled to this, supplying a "virtual battery" which is the same thing. It's just an isolator.

Easy enough to do once you think of it. Tricky to get it to work (a) reliably  and (b) quietly.  But a good candidate for postage-stamp PCBs.

Wow, I can't believe that never crossed my mind before!

For the capacitively isolated version, what were the drawbacks in reliability? What chip did you design it off of?

[MetalGuy]

To save you some trouble below is a very well documented MAX1771 SMSP + PCB:

http://desmith.net/NMdS/Electronics/NixiePSU.html

[Ice-9]

To save you some trouble below is a very well documented MAX1771 SMSP + PCB:

http://desmith.net/NMdS/Electronics/NixiePSU.html

That is the link refered to at the top of the page and is the schematic we have used as a starting point.

[manis404]

Could a variation of this power the tubes in the preamp stage(only the preamp!) of a dual rectifier? It uses 4 12ax7 tubes if i remember correctly.

[frequencycentral]

Could a variation of this power the tubes in the preamp stage(only the preamp!) of a dual rectifier? It uses 4 12ax7 tubes if i remember correctly.

My interest in SMPS's is in powering small amplifiers with them, I've already had success with a similar 555 based SMPS: http://www.diystompboxes.com/smfforum/index.php?topic=78302.0

So, yeah - that's certainly possible, providing the SMPS will provide enough current for 8 triodes. But each 12AX7 plate is something just above 1ma IIR, so I'd say yes, as this SMPS should be able to deliver 50ma.

[MetalGuy]

That is the link refered to at the top of the page and is the schematic we have used as a starting point.

I noticed that now.

...so I'd say yes, as this SMPS should be able to deliver 50ma.

The article says that it can do up to 100mA.

[Salvatore]

Hm, interesting stuff.

(Writes down this trivial comment for future topic notivication. )

[frequencycentral]

Well, I can't justify making an order to Banzai for just a couple of 0R050 and 0R1 1 watt resistors, so this'll have to wait for a little while.....

@ Mick (Ice-9): does the IRF740 in yours get very hot? I ask because the IRF740 I used in my 555 SMPS does get hot and needs heatsinking, and the site the schematic is from states "At no time did any component get hot - the FET was always cool to the touch, and the inductor warmed up only under heavy loads, but never got hot."

[Cliff Schecht]

Well, I can't justify making an order to Banzai for just a couple of 0R050 and 0R1 1 watt resistors, so this'll have to wait for a little while.....

@ Mick (Ice-9): does the IRF740 in yours get very hot? I ask because the IRF740 I used in my 555 SMPS does get hot and needs heatsinking, and the site the schematic is from states "At no time did any component get hot - the FET was always cool to the touch, and the inductor warmed up only under heavy loads, but never got hot."

Rick, if you want I can mail you a few 0.1 Ohm 3 Watt SMD resistors to play around with..