40v 20mA, isolated from 5vdc?

[max_torque]

I need to generate an (approximate) 40v voltage source to drive a 20mA constant current into a burden load, however i need that current loop to be galvanically isolated from the rest of the circuit, that has a 5vdc supply rail

Currently i'm planning on using a 2w isolated 5V:5V DC converter, and then a step up convertor (probably an LT3467).  Cost isn't really an issue, but i want a good level of reliability, so min component count probably helps.

Have i missed some simple way of doing this in one step? I can't (yet) find a suitable off-the-shelf isolated convertor to do it in one stage?

Voltage ripple and stability is fairly un-important because the current loop is set by a CC diode, so anything between around 36 and 46v would do

[BrianHG]

Cost isn't really an issue, but i want a good level of reliability, so min component count probably helps.

Ripoff 42$ direct 5v to 40v, 25ma solution:
https://www.digikey.com/products/en?keywords=%20%09945-2051-5-ND

5$ solution, but, you get 48v, 21ma.  Use a 7-8 volt, 1/2 watt zener diode to drop it down to 40v.
https://www.digikey.com/product-detail/en/cui-inc/PEM1-S5-D24-S/102-3031-5-ND/4006999
21ma out cuts it close.

6.46$ solution, but, you get 48v, 40ma.  Use a 7-8 volt, 1 watt zener diode to drop it down to 40v.
https://www.digikey.com/product-detail/en/cui-inc/PDM2-S5-D24-S/102-3005-5-ND/4006973
This one gives you some breathing room.

[Jeroen3]

Maybe two sets of SN6501 with transformer in series.

[Zero999]

Maybe two sets of SN6501 with transformer in series.

You might be able to get away with a single SN6501, since it's rated to 350mA, which is 1.75W at 5V and he wants 20mA at 40V out, which is 0.8W.

If that's not enough, there are other transformer driver ICs, with a higher current rating.
https://www.mouser.com/ds/2/256/MAX258-257990.pdf
http://cds.linear.com/docs/en/datasheet/3439fs.pdf

[mikerj]

Cost isn't really an issue, but i want a good level of reliability, so min component count probably helps.

Ripoff 42$ direct 5v to 40v, 25ma solution:
https://www.digikey.com/products/en?keywords=%20%09945-2051-5-ND

5$ solution, but, you get 48v, 21ma.  Use a 7-8 volt, 1/2 watt zener diode to drop it down to 40v.
https://www.digikey.com/product-detail/en/cui-inc/PEM1-S5-D24-S/102-3031-5-ND/4006999
21ma out cuts it close.

6.46$ solution, but, you get 48v, 40ma.  Use a 7-8 volt, 1 watt zener diode to drop it down to 40v.
https://www.digikey.com/product-detail/en/cui-inc/PDM2-S5-D24-S/102-3005-5-ND/4006973
This one gives you some breathing room.

Be aware these isolated DC-DC modules are usually quite noisy (electrically) and most always have terrible load regulation.  Apart from that the last one looks like a simple and cheap solution.

[Rog520]

How about something like this?

http://cds.linear.com/docs/en/datasheet/8301f.pdf

[max_torque]

a-ha! I hadn't considered using a +-24v isolated convertor to get 48v!  That should work fine, as the CC diode is rated at 80v (iirc) and under 50v means i don't have to worry about HV (H&S) requirements etc

Now just have to do decide how to get a the analgoue current measurement signal back across the isolated barrier....  (probably with a V to F converter, maybe one of the little TimerBlox ones from LT?)

[David Hess]

I need to generate an (approximate) 40v voltage source to drive a 20mA constant current into a burden load, however i need that current loop to be galvanically isolated from the rest of the circuit, that has a 5vdc supply rail

Cost isn't really an issue, but i want a good level of reliability, so min component count probably helps.

Have i missed some simple way of doing this in one step? I can't (yet) find a suitable off-the-shelf isolated convertor to do it in one stage?

Voltage ripple and stability is fairly un-important because the current loop is set by a CC diode, so anything between around 36 and 46v would do

Use an 1:8 transformer (or two 1:4 transformers) as part of a high frequency inverter.  This is simple and especially so for the level of performance available.

Pulse and gate drive transformers are suitable.  A quick search did not turn up any with exactly the right turns ratio but do not forget that a center tap can be used to double or halve the turns ratio.

[BrianHG]

a-ha! I hadn't considered using a +-24v isolated convertor to get 48v!  That should work fine, as the CC diode is rated at 80v (iirc) and under 50v means i don't have to worry about HV (H&S) requirements etc

Now just have to do decide how to get a the analgoue current measurement signal back across the isolated barrier....  (probably with a V to F converter, maybe one of the little TimerBlox ones from LT?)

http://www.analog.com/en/products/analog-to-digital-converters/precision-adc-20msps/isolated-ad-converters.html
http://www.ti.com/product/amc1106m05
https://www.maximintegrated.com/en/products/sensors-and-sensor-interface/MAX14002.html  This guy is lower res, costs more, but, has a built in DC-DC converter...
https://www.maximintegrated.com/en/products/sensors-and-sensor-interface/MAX14001.html  This guy is lower res, costs more, but, has a built in DC-DC converter...

[mariush]

HORRIBLE efficiency.. (<33%) but you could put 2x adum5000 in parallel to get up to 5v 1w and then boost to 40v

[BrianHG]

HORRIBLE efficiency.. (<33%) but you could put 2x adum5000 in parallel to get up to 5v 1w and then boost to 40v

What are you talking about?

[mariush]

I'm talking about Adum5000 chips, they're digital power isolators which can do up to 5v at 100mA (500mW) but the efficiency is poor ... 34-35% :

See page 5 , 34% max at 100mA output : http://www.analog.com/media/en/technical-documentation/data-sheets/ADuM5000.pdf

So he could have two in parallel and have 5v 600mA in , 5v 200mA out (1000mW), and then a boost regulator could produce 40v at 20mA (800mW) with around 85% efficiency maybe.

So yeah, it would be very tiny, low height, but expensive and low efficiency.  You'd save some money on a potentially custom high frequency transformer for isolation purposes, but you'd spend the money on the adum chips and you'd risk not being able to buy them...

[BrianHG]

I'm talking about Adum5000 chips, they're digital power isolators which can do up to 5v at 100mA (500mW) but the efficiency is poor ... 34-35% :

See page 5 , 34% max at 100mA output : http://www.analog.com/media/en/technical-documentation/data-sheets/ADuM5000.pdf

So he could have two in parallel and have 5v 600mA in , 5v 200mA out (1000mW), and then a boost regulator could produce 40v at 20mA (800mW) with around 85% efficiency maybe.

So yeah, it would be very tiny, low height, but expensive and low efficiency.  You'd save some money on a potentially custom high frequency transformer for isolation purposes, but you'd spend the money on the adum chips and you'd risk not being able to buy them...

Ok, noted for potential future projects where a flat size is needed.
The 6$-15$ isolated DC-DC converters are already isolated, no xformer and you get the 48v, or any voltage from 3.3,5,6,10,12,15,24,30,48, @ 1watt,2watt,3watt(getting expensive),5w(getting very expensive) directly in 1 step with around 80% efficiency, but, it's a small brick shaped device.

I don't know why people complain about noise, I usually place a Polymer Aluminum low ESR cap at the input and output, an they become clean as a whistle.  Sure, just placing just the recommended electrolytic cap in the data sheet leaves you with the ripple noise listed in the data sheet.  But, the data sheet doesn't say you cannot use caps with around 2 orders of magnitude smaller ESR to make to output clean as a battery.

[ikrase]

It just so happens that I also need something like this -- only I need the voltage to be variable and to be tightly/accurately controlled -- and it's going to be summed with some hundreds-of-volts RF (blocked by 2 mH chokes).


I also don't need much current at all - sub-milliamperes of leakage probably.

Should I just have an entire floating microcontroller + DAC + high voltage Op-Amp system, all controlled over optoisolators or isolated RS485 or similar? Or would a better plan be to use so-called "isolation amplifiers"?

[BrianHG]

Just read my links.  It's obvious is isolated ADCs exist, the isolated DACs exist as well.
If you need good regulation, just overshoot on the voltage and use a linear regulator.

As for isolated direct to 200v, yes, those DC-DC converters exist as well, like the first 42$ unit I listed.
https://www.digikey.com/product-detail/en/recom-power/R12-150B/945-2053-5-ND/3776801
This particular one is output voltage adjustable via trimpot, or, DAC and it is far more precision.  Goes from 92v to 210v at 50ma.  Read the data sheets.  Though, it needs 12vdc input.  If you use their 15vdc in or 24vdc input models, you will waste less power.