Power supply sequence

Hi All, I've got this power supply that is a bunch of CUI dc-dc converters. (+24, +15, -15, +5, -5) all made from a 24V wall wart. There's a limit on the input to output voltage on the CUI's so had to make the -15 from the +15.

formatting link
(aside) Hmm that doesn't list limits on voltage for negative Vout. This one does.
formatting link
I guess I should contact CUI and ask them. (end aside)

Anyway because of the sequencing the +15V comes up 1 ms before the -15V. This was fine until a colleague tried to use it to run one of his circuits. It pukes when turned on and the -15V never get's there. (it sits at ~ +0.4 V) (I still have to look into his circuit.) I was thinking of switching the +15 so it only came on (at the output) with the -15. Below is my first crack at a circuit. (PFet is switched on by pnp that gets base drive from +15 -> 20V zener -> -15.)

So is this kosher? Is there some better way? I haven't thought that much about resistor values. TIA

George H. (hmm I could get rid of the pnp and just have the zener turn it on.)

Version 4 SHEET 1 880 680 WIRE -96 -64 -320 -64 WIRE 96 -64 -96 -64 WIRE 176 -64 96 -64 WIRE 368 -64 272 -64 WIRE -320 -48 -320 -64 WIRE 368 -16 368 -64 WIRE 96 48 96 16 WIRE 192 48 192 -16 WIRE 192 48 96 48 WIRE -320 80 -320 32 WIRE -240 80 -320 80 WIRE 96 80 96 48 WIRE -240 112 -240 80 WIRE 368 112 368 64 WIRE -320 160 -320 80 WIRE 96 224 96 160 WIRE -96 272 -96 16 WIRE -48 272 -96 272 WIRE -96 320 -96 272 WIRE -320 400 -320 240 WIRE -96 400 -96 384 WIRE -96 400 -320 400 WIRE 96 400 96 320 WIRE 96 400 -96 400 FLAG -240 112 0 FLAG 368 112 0 SYMBOL pmos 272 -16 M270 SYMATTR InstName M1 SYMATTR Value RSJ250P10 SYMBOL pnp 32 320 M180 SYMATTR InstName Q1 SYMBOL zener -80 384 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D1 SYMATTR Value PTZ20B SYMBOL voltage -320 -64 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value +15 SYMBOL voltage -320 144 R0 WINDOW 3 -332 50 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V2 SYMATTR Value PULSE(0 15 10m 1u 0 1000m 1 1) SYMBOL res 80 -80 R0 SYMATTR InstName R1 SYMATTR Value 10k SYMBOL res 80 64 R0 SYMATTR InstName R2 SYMATTR Value 10k SYMBOL res -112 -80 R0 SYMATTR InstName R3 SYMATTR Value 10k SYMBOL res 48 256 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R4 SYMATTR Value 10k SYMBOL res 352 -32 R0 SYMATTR InstName R5 SYMATTR Value 1k TEXT -354 410 Left 2 !.tran 0 20m 0 TEXT -56 344 Left 2 ;20V zener

Reply to
George Herold
Loading thread data ...

Full schematic at...

Message-ID:

Should be OK. I agree that the PNP is redundant. The only issue may be lack of hysteresis.

Having been burned once on a dual supply chip that locked up with the wrong power supply sequence, my standard procedure is to test/simulate both ramp-up and sequencing and watch for any abnormalities.

Sometimes I build-in power steering elements to avoid any ambiguities. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

Thanks Jim, It works fine in practice too. (no pnp)

I didn't look hard, but there were some "squirrelies" on the

+15V source line as it turned on... nothing on the output (drain). Is that perhaps your hysteresis warning.

George H.

Reply to
George Herold

Hysteresis would ensure that the pass device doesn't spend too much time in the active region. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

As a chip designer I'm a belt and suspenders type... otherwise chips fail, so this is what I came up with...

I tried the Rohm website model of your RSJ250P10 device... it seems very leaky and/or has a huge Qgate, which may explain what you were seeing on the +15V rail. If that is your device of choice I'd add a PNP to drive the gate. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

Wow, Thanks Jim. So you added a hysteresis part. I don't really see why this is needed, but I appreciate the work! It switches fine right now... (With 50 ohm loads on the outputs I do see a few volt drop in the +15 as the load is turned on.. for a few milli-seconds.

The metal box for the board is 1/2 designed, and I like the keep the pcb the same size that it is. (Other wise more work) It looks like I can fit the parts in. (I'm also adding layout FET's for the +5V and -5V, won't be populated... just in case.) (Hmm will I need one for the +24V too...?)

I've also written to CUI, asking about maximum voltage on their part.. but I haven't heard back.. maybe the whole circuit will be moot.

No I'm not using the Rohm device.. it was all I could find in LTspice that sorta matched what I'm using. (IRF5305, our stock P-fet)

Thanks again, George H.

Reply to
George Herold

On Thursday, January 26, 2017 at 11:36:53 AM UTC-5, George Herold wrote: < snip another incomplete circuit description> The -15V is most likely not coming up due to overcurrent. Quite a few of th e older ICs requiring bipolar supplies do not like unipolar one bit especia lly if input networks go outside the tolerable CMR with an out of spec appl ied power supply. You might want to take a closer look at your load before you blow something.

Reply to
bloggs.fredbloggs.fred

Some linear vregs won't start up if you drag their outputs the wrong side of ground. A Schottky diode from each supply to ground is always a good idea, and IME eliminates most sequencing worries in analogue stuff.

Cheers

Phil Hobbs

Reply to
pcdhobbs

Some parts, linear and mixed, will latch up if a rail transiently blips to the wrong sign. Any part that has pins V+ and V- and GND is a potential victim.

Absolutely.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

If you have the room, and if there's any danger that power and ground may be reversed on a board, put in a fuse followed by said Schottky. It holds down on the carnage on the manufacturing floor.

--
Tim Wescott 
Control systems, embedded software and circuit design 
 Click to see the full signature
Reply to
Tim Wescott

I can post a schematic of the power supply. Vin, three terminal voltage regulators (switch mode), capacitors (10 uF ceramic, "50V") some ferrite beads, and more caps.

I don't know about the load, but it has a power opamap, that can "out muscle" the power supply current-wise.

George H.

Reply to
George Herold

I usually use a polyswitch and unidirectional TVS on the input supply, which is generally good enough. I really like Tyco PolyZens because they dump heat from the zener into the polyswitch, which makes the whole thing faster and more reliable.

Polyzens behave almost like voltage regulators at light load. I sometimes put a TVS in series with the ground lead to get higher trip voltage, but then a supply reversal will pull two diode drops below ground, which is a bit nervous-making, so an extra Schottky is sensible.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Unless it unsolders itself and falls off. I have on my shelf a fake Apple USB wart that runs a little Royer switcher full blast into a 5.1V

1W zener.

No isolation slots, minimal creepage distances, you know the deal. The Zener unsoldered itself from the board, but still tests good.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

We are currently having that debate. I often use a polyfuse and a transzorb on a box input, usually DC from a wart. Customers might use their own supplies, so we can't depend on the wart to limit current, or even get the polarity right.

People are pointing out that one can tease the polyfuse+transzorb input voltage and fry the transzorb. True, but that's improbable, especially for 24 volt inputs, where someone applying a higher voltage is unlikely.

If the transzorb fries, it shorts, and still limits overall damage.

There are some new IC fuses, so maybe we'll start using them.

formatting link

Nice, but complex.

The polyzens look better.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Thanks for the idea Phil. It didn't work in this case. In fault the -15 V is sitting at ~ (+)280 mV. not enough to turn on a Schottky. I sent an email to CUI last week, but have heard nothing.

I tried to design something to sequence the -5V and +5V too.

+5V was fine. But the -5 was a fail. I kinda stink at transistor level design stuff. I have trouble with thinking about things with p-channel. (pnp's) It's all upside down. I ended up doing it with N-fets and npn's and then flipping it all over*.

(For the -5V the -15V rail is at ~gnd, and this turned on the pnp "backwards" current flowing the wrong way. I couldn't really "see it" till I flipped it over.)

George H.

*I totally blame it on AoE, and their npn, N-fet bias, I have to blame someone after all.. can't be my fault. (tongue is firmly in cheek. :^)
Reply to
George Herold

I had trouble getting the polyfuse/ zener idea work for higher voltages, (and reasonable currents.) As Phil said, 5V was fine, and with low input voltages the zener was happy doing it's thing as a voltage regulator.

George H.

Reply to
George Herold

OK final follow up to my thread. I got word from CUI, and I can supply the

-15 from the +24V! Life is good, (for the moment. :^)

George H.

Reply to
George Herold

Grumble, OK one final update... in case there is someone following along at home. Even when driven from the same input the -15V came up later (~0.5ms) and hung the circuit. Turns out it wasn't the power opamp that was sending voltage down the negative rail, but "normal" opamps. There are five opa2134's in the circuit. Un-powering any one of them let's the circuit work. Since it's low current opamps I made an RC (10 ohms 100uF) on the positive rail feeding the opamps. It works, but it's a band aid.

George H.

Reply to
George Herold

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.