Using a DC-DC Isolator to allow single power input to case?

Question

I'm working on something where I use an Arduino talking via I2C to a DAC, an ADC and an RTC and I'm provisioning an I2C isolator and a digital isolator (for digital pins) to provide separation between the two. I'm doing this to prevent blowing up the Arduino if something goes wrong and also just because I want to as part of experimenting/learning.

This only works if both power and ground on sides A and B of the isolators are separate from each other. To complicate matters, if this all works and I encase it then I only want one power cord into the case - something like a 15V wallwart affair. This would, inevitably, tie grounds together and the isolators won't work: not 'won't work in an isolation' sense, but not work at all - I've tried it!

So I'm wondering if I could feed the 15V supply into a 12V regulator and subsequently a 5V regulator for Side B (I need 12V and 5V on that side) and into a 12V DC-DC isolator that outputs 5V for side A which also powers the Arduino.

Something like this Tracopower supply. When I look at this I see that the Vin ground is connected on Side B, but the Vout -Ve is isolated for Side A so it seems like it would do the trick. I thought I'd ask here first before committing to spending money to try it out or indeed if there were other ideas.

My main criteria are:

5V, 1A output

Clean enough to power an Arduino and 5V vin isolator (I could use a supply with a higher ouput voltage and feed through a 5V regulator)

Only 1 main power input. If there's no solution that allows for this, I'd do away with the isolation altogether. In reality it's not really needed.

jw0752 · Accepted Answer

Hi Andrew,

Yes this is an isolated DC to DC converter and it will provide a voltage that is not common to the input voltage. this should allow you to do what you are talking about though I am not weighing in on that. Many DC to DC converters are not isolated so make sure you get one that is designed to be. These little DC to DC converters are very efficient and very useful. A company called Recom also has a good line of converters.

https://www.newark.com/b/recom-power?ost=recom&rd=recom

John

jw0752 · Answer

Hi Jan, I believe there were two models listed on the data sheet. One had +/- with a common and the other was just +/- with no common. If I read it correctly both of these options were isolated from the Input ground. John

Jan Cumps · Answer

If you need single output, you're good. The output is isolated. If you want me to pre-check something, I have one of the same family;

Jan Cumps · Answer

I've already blown one £76 instance when I was building my power supply through inadvertent probing. That's expensive! By coincidence, today, balearicdynamics and I had a chat on preventing this When I use my scope to measure situations where I want to put the probe reference on a different point than Ground, I use this: They aren't cheap - I believe that Gough Lui has fou nd a good cheaper one, if I recall correctly. On the other hand, they prevent damage to your scope or device under test when you need to probe without ground reference - or without short-cutting a circuit node to ground if you accidentally put your crocodile clip on a non-ground node.

Jan Cumps · Answer

i believe that the two outputs aren't isolated from each other. It's balanced out: Vmin - Common - Vplus.

dougw · Answer

TEL 5-1211 looks adequate to supply 1A max at 5V with 1500 V of isolation to both output pins. It will dissipate over 1 W of thermal power at 1 A output, so it will get pretty warm.

michaelkellett · Answer

I've used lots of Traco DC-DC converters and found them to be reliable but quite noisy (no worse than other makes at similar prices) so output filtering may be helpful. The other issue is that the start up current of the Traco converters can be quite high, so while you may have a run current of 150mA you might need 700mA to get it to start - worth testing in a lashup if you can. MK

Jan Cumps · Answer

This discussion has pushed me to do something I was planning since some time: make a power supply for my eload. I'm a known procrastinator. I need a push to do somethng that's not on my immediate horizon. This thread made that happen . I'm turning the evaluation board of the Infineon Gate Driver with Truly Differential Input - Review into the supply for the analog part. michaelkellett wrote: The other issue is that the start up current of the Traco converters can be quite high, so while you may have a run current of 150mA you might need 700mA to get it to start - worth testing in a lashup if you can. MK I will test that, Michael, and share the results here ...

Andrew J · Answer

The isolator arrived and I hooked it up - in the end I went for a Recom RS6-1205S . It's very similar to the Traco one, a bit cheaper. It actually works for what I want to do: using it to power Side A of the i2c and digital isolators as well as the Arduino. I did some quick measurements: Initially load was just a 1K resistor: AC with 0.1uf and 1uf capacitor on the output: DC with 0.1uf and 1uf capacitor on the output: AC with 0.1uf, 1uf and 10uf (electrolytic) capacitor on the output: DC with 0.1uf, 1uf and 10uf (electrolytic) capacitor on the output: Powering side A, still with 0.1uf, 1uf and 10uf (electrolytic) capacitor on the output: AC: DC: It's quite noisy, as Michael alluded to, and there is quite a large amount of ripple induced on the DC (datasheet states 25mV to 75mV) but it's on a breadboard without any great output filtering. I still need to work through any impact this has on Side b (where the DAC and ADC are located.) In theory, I would expect none as they are driven through I2C commands and the DAC is powered from a REF5040 which is used as the reference source for the DAC output.

Jan Cumps · Answer

When I put an I2C isolator in between with both sides receiving power independently from two different regulators BUT with a common ground it wouldn't work: SCL was fine, but SDA just couldn't get pulled low. Once I switched to powering Side A of the isolator from the Arduino (USB power) with its ground, and Side B from the regulator it all worked ok. In fairness to me that was what I was expecting but it doesn't hurt to test these things out. I've used an isolator (TI ISO1541) in a context with and without GND shared ( I created a PCB with possibility to break ground ). It worked in both cases. Your scenario makes sense. Powering the left side of the isolated circuit from the adapter, then using that same adapter to power the input of the TRACO. The isolated outputs of the traco can then power the right side of your isolated circuit. Little disclaimer: the Traco's isolation is called "functional isolation". It's not the same as safety isolation. It'll do the same but you can't complain to Traco if you die . edit: The isolator is sensitive for correct pull-up values. As you can see in my scope trace here, in the light blue trace, there are situations where the SDA low is higher than expected. That's typical in an isolator scenario. shabaz , you also have experience with this I believe. Do you have advise for Andrew on using an isolator for i2c with both sides sharing power (or ground) - although that defeats the need of an i2c isolator ?

Gough Lui · Answer

Indeed, I tested both the HVP-70 and Micsig DP10013 in this posting: https://goughlui.com/2018/12/16/review-eevblog-hvp-70-70mhz-micsig-dp10013-100mhz-high-voltage-differential-probes/ You do get what you pay for to some extent ... if you need isolation for high-voltage circuit probing, these are good options. But if you're interested in low voltage, then you might want a standard differential probe otherwise the noise might cause problems for your measurement. As for the original query - yes, many Traco Power modules are isolated, and this is the approach taken by the Infineon Differential Gate Driver demo board as well as the B&K Precision DAS240 Multi-Channel Logger (along with a digital isolator) to provide isolation. - Gough

Andrew J · Answer

I think I have just confused things not actually posting in detail but I thought that would just cloud things in respect to my question! Obviously mistaken!! The grounds don't need to be connected, it wasn't a specific requirement. In an ideal isolated world they wouldn't be. What I wanted was to try and protect the Arduino, not to have to plug it into a PC to power it, and not to have two input power supplies - a 15V and 5V as drawn on the right in the image in the post above. I have components that require regulated 5V and regulated 12V and the 12V regulator requires a minimum 15V input. I can't power Side A from the Arduino once encased because the Arduino needs powering and I don't want to plug it into a PC or 5V wall wart (back to two sockets!) My first stab at that, because I didn't know any better at the time, was to isolate just the power and signal traces and not to worry about the ground as I drew above. So the grounds are the same purely from initial expediency - saves some money as galvanic isolation isn't really required. That doesn't work - the I2C isolator doesn't seem to like it (perhaps just because it's on a breadboard??) - so I need to move to my next idea which is to use a DC/DC isolator. I can use either the 15V or 12V as input to the DC/DC isolator (12V is its nominal voltage.) Now in terms of 'saving money' I was using a LM7805 as one of the regulators, which I have to hand. That doesn't like powering the Arduino with a voltage drop from 12V to 5V - as I expected, gets too hot and goes into thermal shutdown eventually. Given that I'd need something like the Tracopower 5V switching regulator - the other one on my image - at around £5 - I might as well not worry about saving £10 and just get the DC/DC isolator and get proper isolation. So my question was really just making sure I was thinking straight, before I spent that £15 to test it out. I probably provided more information than necessary and just confused everyone. Still, it's led to another interesting discussion! (I forgot to draw on the 5V regulator on the right hand side where it is marked 5Vdd - imagine this is a 5V regulator!) This then allows me to have one plug in the wall, no connection to a PC and drive both the Arduino and power components completely isolated from each other. Again, I haven't put in all the caps, resistors, diodes etc.

Jan Cumps · Answer

Input current, power supply without current limit, small load on the +12V. -12V open Traco Power TEN 64822 Input 48 V: Input 24 V (input current when stable: 20 mA): Almost max load on +12 V, -12 unloaded: (45 Ohm , 245 mA, input current when stable: 160 mA)

Andrew J · Answer

Given this is on a breadboard at the moment there's likely to be some noise. Side A: Arduino providing I2C comms to the isolator. On this side is also power to an encoder and power button. Side B: I2C comms to DAC, ADC and RTC. There's also a 5040 voltage ref. Here's a pic (I missed out pullup resistors, caps, bypass and reverse voltage diodes etc to avoid cluttering): (NOTE: on the image above, I've shown Vin and GND to the Arduino which would be the case when not connected to the PC during testing/development. When I was trying this on the breadboard, the Arduino was powered over USB from my PC) What I have above doesn't isolate at GND - but avoids having two power inputs (what I have drawn on the right) and I thought it might be 'good enough' if I tied the grounds under the trace to the left-hand 5V regulator. As I mentioned I don't NEED isolation I was just doing it out of interest as much as anything. I know Jan/Peter did this with their DC load to prevent any high voltage/current damaging the USB port of the connected computer. They can provide complete isolation on both sides because one is powered by USB and one by a 12V supply; these can be bridged if the MCU isn't connected so both sides are powered by the 12V supply. I absolutely didn't want to drive this with two separate power supplies either from a PC or, say, a 5V wall wart on top of the 15V wall wart and the Arduino is always connected. With the above setup (Arduino powered from USB), when probing the SCL line on the isolator Side B it was rising to +5V/falling to 0V as expected; however the SDA line on Side B was rising to +5V and was only pulling down a few hundred mV below 5V and in a very, very non-square wave form, with very occasional pull downs to 0V. Trying different pullup resistors had only a minor impact on this. When I did away with the Side A 5V regulator and took Side A 5V and GND from the Arduino, it all worked fine. So, my first 'good enough' stab at this won't work but if I swap the 5V reg on Side A for the DC/DC isolator it seems like it should achieve what I want.

Jan Cumps · Answer

Hot plugging the input, with the power supply already on 24V. +12 V loaded to 250 mA, -12V open. Is this the inrush you refered to, michaelkellett ? edit: When I restrict my PSU to 400 mA, then hot plug the Traco, it starts just fine without peaking to 1.4 A. I'll check what the minimum current it needs to get started... edit edit: My PSU current restriction isn't fast enough. When I put it on 200 mA, the inrush still peaks at 1.025 A during 2 samples

Jan Cumps · Answer

1.4 A is because I put my meter in the 1 A range. 1.4 is the limit then. If I put it in the 10 A range, it also peaks out. That's normal. The input capacitor is a shortcut initially. Only the PSU impedance, the test leads and the capacitor's ESR restrict the inrush. it's only a few samples wide. There's virtually no energy there. Every hard-switched DC circuit with an input capacitor has this phenomena. The capacitor in my circuit is 47uF. Input capacitors cause inrush. The more capacitance, the higher the inrush energy. 220 uF would increase the (not really a) drama... edit: you don't have to worry about it. You just need to take care that you have enough oooomph in the supply to get going and deliver the nominal current once stable. It would only be an issue if your source collapses under the initial demand and doesn't recover. Here's what the output switch-on curve looks like, with 125 mA current (96 Ohm load):

Using a DC-DC Isolator to allow single power input to case?

Top Replies