EEVBlog #285 - Isolated USB Lab Supply

[JoannaK]

STOP shifting the project into the weirdest areas :
1) Cheap simple and not complicated
2) Isolated

All he wanted was a simple Isolated USB Lab Power Supply without all the added bonuses like microc's and such

Let me ask you, does your wing-wang-pong power supply have a microc monitoring system? Yes?

If someone wants to design/make decent adjustable power supply with settings and measured values shown .. IMHO it's better be microcontroller based.. I know One could do it with pure amalog circuits, 4000-series logic or code it all on smallish FPGA, but in the end that's system cheap 4-8-16 bit µC is quite suitable.

china-Cheap producs are usually made with as old and as cheap parts as can be found. Usually even cloning decades old designs and layouts.

[chrome]

STOP shifting the project into the weirdest areas :
1) Cheap simple and not complicated
2) Isolated

All he wanted was a simple Isolated USB Lab Power Supply without all the added bonuses like microc's and such

Let me ask you, does your wing-wang-pong power supply have a microc monitoring system? Yes?

WRONG, he does have a µC, how else do you think he controls the Display or read the rotary encoders...

[T4P]

NO. Like a microc talking to a computer? That was what i was talking about.

[chrome]

It's 3 part more, he doesn't even have to add them on the PCB, just add the footprints (FTDI chip and 2 optocouplers).

[EEVblog]

It's 3 part more, he doesn't even have to add them on the PCB, just add the footprints (FTDI chip and 2 optocouplers).

Correct, I don't have to add them.
I am considering putting in the footprints if I have room.
FYI, the FTDI chip then becomes the most expensive chip on the board, by double.
Still cheap, as chips go, but double...
And my micro doesn't have a UART, so it would have to be bit banged.

Dave.

[PA3BNX]

Hello EveryBody,


I do not need a USB powered supply I first thought.

But maybe it's nice to feed my simple SuperSimpleDoppler (RDF) circuits.

In one box a USB Soundcard device with my Doppler driver circuit
all feed through  the Laptop USB port power.

That would be nice for me and maybe many other fox hunters with
with there laptops fox hunting mobile.

Maybe it needs Audio isolation
to the USB Soundcard and Circuit to make sure no destroying loops
could occur from the car battery then.

So your eevblog did make me think again.

I know this is off topic here hi.

[juanfermed]

Hi everyone! Nice conversation about this USB powered PSU. I read the 11 pages of the conversation and maybe I missed it but: what are the specifications of this PSU? I know it will be USB isolated (a lot of comments on that) but what about current it will be able to provide and the voltage? I know that it is still on design and has to be decided but does anyone knows if Dave has decided how much voltage and current he wants this PSU be able tu supply?

Thanks!

Juan

[hlavac]

Looking at the input power available and what Dave said about the architecture:

With the isolated DC-DC conversion, linear regulator in tandem with variable voltage DC-DC boost preregulator and LED displays, and output side power rail for powering a microcontroller which would also have to magically appear at the isolated output side where the microcontroller would be, and about 4.2V/500mA worst case input power due to USB cable voltage drop, my estimates come to at best 1W reliably usable for output.
That is 3.3V/303mA, 5V/200mA, 10V/100mA, 12V/83mA, 15V/66mA, 20V/50mA.

If you want more you would probably have to drop something.
Some power could be saved on the LED displays. Assuming multiplexed digit mode, up to 8 segments lit with 20mA each at any time, and classic resistor current limiting on these from 5V power rail, LED displays alone can take up to 0.8W. This could be lowered to about 0.4W if powered from a low voltage supply rail right above the LED voltage drop + digit switch drop. You could also reduce the current/pwm on ratio at the cost of reduced brightness. And get some high efficiency LED displays that need much less than 20mA to light up probably at higher price.

Drop the linear regulator and leave just isolated variable DC-DC with an output LC filter for about 20% more power in exchange for a bit of output ripple.
Drop LED displays and replace with LCD with efficient single LED or no backlight for another 10% or so.

Maybe you could get down to 0.5W overhead/1.5W usable supply for
3.3V/454mA, 5V/300mA, 10V/150mA, 12V/125mA, 15V/100mA, 20V/75mA.

Not sure how much lower you can get, with classic optocouplers for isolation probably not as these include power hungry IR LED that will easily want 5mA, 6.5mA if you want to leave slack for LED burnout over time... Maybe with some fancy inductive type of isolators? I have even seen an isolated op-amp with these somewhere...

I can't wait to see how efficient will Dave get with this

If he changes his mind and actually does use some USB talking chip, a great feature to add would be an isolated built in logic level RS-232 port for debugging the powered device

[T4P]

I saw a isolated opamp on EEWeb, that is, for high voltages

[EEVblog]

I can't wait to see how efficient will Dave get with this

Not very. Two stages of about 80%

Dave.

[SgtRock]

Greetings EEVBees:

--Great thread. I learned a lot.

--I agree with Dave. Small and simple, is the way to go. Being that USB is ubiquitous, this device will be kind of like the Swiss Army Knife of Power Supplies. One can always go get a Ka-Bar, if real mayhem is called for.

--The horizontal binding posts are a great idea. Shortening the height, and getting rid of the empty space in the box is another problem entirely. Please see the attached pictures, where I cut down a standard project box (one off) with a razor knife, to house and shield the Sparkfun Cap Meter. I do not quite know how DJ could achieve this in a practical manner. No doubt, it is not cost effective to order a custom box, that is only as tall as it needs to be. Perhaps someone has an idea.

--I was also glad to see that people could disagree, without being disagreeable.

"All I wants of you, Cap'n Simmons, is plain seevility, and that of the commonest goddamndest kind!"
Zeph W. Pease

Best Regards
Clear Ether

[Psi]

It's 3 part more, he doesn't even have to add them on the PCB, just add the footprints (FTDI chip and 2 optocouplers).

Correct, I don't have to add them.
I am considering putting in the footprints if I have room.
FYI, the FTDI chip then becomes the most expensive chip on the board, by double.
Still cheap, as chips go, but double...
And my micro doesn't have a UART, so it would have to be bit banged.
Dave.

Doesn't have to be the $6 FT232RL, there are a few cheaper ones for about half that.
Though i guess if its left unpopulated on the pcb it's better if it's something common.

[EEVblog]

It's 3 part more, he doesn't even have to add them on the PCB, just add the footprints (FTDI chip and 2 optocouplers).

Correct, I don't have to add them.
I am considering putting in the footprints if I have room.
FYI, the FTDI chip then becomes the most expensive chip on the board, by double.
Still cheap, as chips go, but double...
And my micro doesn't have a UART, so it would have to be bit banged.
Dave.

Doesn't have to be the $6 FT232RL, there are a few cheaper ones for about half that.
Though i guess if its left unpopulated on the pcb it's better if it's something common.

I'm going by the 1K Digikey price for the FT230XR-S, and stand by my original statement!

Dave.

[lindworm]

For isolating the FT23* you can use a nice little ADUM 1201, they are quiet cheap, don't require external components exept two 100n bypass caps and are much better suited than most optocouplers.

[nitro2k01]

I'm still worried about the binding posts. Sure, they'll probably fit, but how is the durability? Don't want the damn things to come loose after 6 months. And if you're going to use taller buttons, you'll still need to drill holes in the clear cover. One option might be to use a PCB as the side panel on the USB side. This way you could put the buttons on that board, and not use elevated buttons.

[FenderBender]

http://www.digikey.com/product-detail/en/6883/501-1316-ND/737883

I think this would be a good jack to use. Digikey per 1000 prices are $1.84 each and that's for both positive and negative. Plus it's lower profile than others and hey made in the USA if that matters.

[amspire]

http://www.digikey.com/product-detail/en/6883/501-1316-ND/737883

I think this would be a good jack to use. Digikey per 1000 prices are $1.84 each and that's for both positive and negative. Plus it's lower profile than others and hey made in the USA if that matters.

Not sure if there is anything low profile about those sockets. 50mm long and they project over 30mm front the panel. They would take more room then the cheap sockets Dave showed us - I doubt they would even fit.

[T4P]

http://www.digikey.com/product-detail/en/6883/501-1316-ND/737883

I think this would be a good jack to use. Digikey per 1000 prices are $1.84 each and that's for both positive and negative. Plus it's lower profile than others and hey made in the USA if that matters.

Pricey and far too long.
Remember, cheapie power supply without manufacturing the whole thing in china but with the cheapest components as possible

[gxti]

I'm going by the 1K Digikey price for the FT230XR-S, and stand by my original statement!

MCP2200 is cheaper even after the required external crystal, 1.47 in hundreds and 2.12 in singles. Direct from manufacturer might be different but I'm too lazy to look it up. Not sure why everyone is so fond of the FTDI parts for trivial UART applications.

But it's not up to me to tell you how to bloat your projects. Maybe you should have chosen something that everybody doesn't think they understand :-)