OT: boost converter questions

newguy · Joined: 24 Jun 2004 Posts: 1903

I have a pet project that runs from 12Vdc. I use one particular sensor that will run off of a 18-30V supply. The thing will actually function at 12V, but not its analog output. This output is a 4-20mA or 0-10V analog output (switchable). I need the 0-10V analog output to work, so I have to supply it with at least 18V.

For my first stab at the prototype, I just ran the thing from 2 x 12V supplies in series (actually batteries). That works, but I want to get rid of the second battery. That's why I'm considering a boost switching converter.

The National Semiconductor simple switcher application will create sample designs easily enough, but the output ripple is a concern as I don't know how the sensor will react to ripple and I don't have the equipment to simulate or superimpose ripple onto my supply to see how it behaves. There's something else that complicates things: I don't run this sensor all the time. I turn it on & off quite randomly. I know that switching power supplies need a load to run properly, so I'm considering the use of a high side switch to turn on/off the switcher, not the sensor. In this way, the switcher won't ever see the load disappear. However, this leads to startup issues (or does it?)

The last thing I'm worried about is switching noise being backfed through the 12V supply while the switcher is on. Can I add filtering upstream of the switcher without affecting the switcher's performance? For that matter, can I add filtering downstream of the switcher (I think the answer to this question is no, but I'd appreciate confirmation/denial).

Thanks in advance.

languer · Joined: 09 Jan 2004 Posts: 144 Location: USA

If your current consumption is not too great, you could consider a charge pump voltage multiplier (MAXIM and LTC have some parts for this). The voltage ratios for this type devices are 1.5X or 2X.

However you mention the output is switchable from 4-20mA to 0-10V. If you can use the current output, you could use transimpedance amplifier which will not require any boost of output.

newguy · Joined: 24 Jun 2004 Posts: 1903

Thanks for tips, I'll look into them in the morning. Regarding using 4-20mA instead of the 10V out, that's not an option. The device has 4 different configurations and the two with the 4-20 current loop output or 10V voltage out require a minimum 18V supply to work properly. I did try it in I loop mode, but the output doesn't behave properly with supply voltages less than about 17V.

Ttelmah · Guest

Have a look at Maxim application note 2027.
Unfortunately, the IC's they are discussing, won't give the sort of voltage you want, but the basic approach will work fine. So consider a simple LDO linear reg, giving (say) 10v, feeding one of the National parts. Quite a few of the National parts have shutdown input pins, which makes turning them on/off easy. Yes, you will have to turn the unit on, and wait for a small time for the output to stabilise (in fact I'd expect the sensor to have a similar, and possibly worse requirement in this regard...).
Also, if the sensor is noise sensitive, consider a second LDO, on the output of the switcher feeding this. However good output filtration may be all that is needed.

Best Wishes

newguy · Joined: 24 Jun 2004 Posts: 1903

Thanks. I was actually considering this kind of approach to isolate the switcher from everything else. The sensor itself takes 300ms to stabilise and start spitting out valid data. I expecet that the few microseconds that the switcher takes to stabilise shouldn't have much of an effect on things.

newguy · Joined: 24 Jun 2004 Posts: 1903

Just for the record in case anyone else encounters a problem like this, here are some preliminary findings.

I ordered a National Instruments simple switcher evaluation kit. I'm using a LM2585S-ADJ to create a 24V, 0.6A supply from 12V. I'm happy to find that it doesn't care if a load is connected or not - it always outputs 24V no matter what. I was worried about this in particular.

Upstream noise (noise thrown onto the input 12V rail from the switching supply) is pretty significant. -0.2V +0.075V spikes on the supply at the switching frequency of 100kHz. I'll have to filter it for sure.

Downstream noise (on the 24V rail created by the switching supply) isn't so bad. +/- 0.1V @ 100kHz, but I'll still have to filter it. I tried doing a quick experiment where the 24V feeds an ordinary 7812 linear regulator (12V). There are still spikes on its output, but their amplitude is about +/- 40mV.

I'll do some more experiments in the morning with the sensor I mentioned earlier. Hopefully I can get away without filtering for it, but I'll definitely require filtering to isolate the switcher from my 12V feed. I have too much sensitive analog circuitry that I can't afford to corrupt with noise.

newguy · Joined: 24 Jun 2004 Posts: 1903

To close this topic for anyone who happens to do a search in future......

I made sure that the boost inverter was fed by an upstream LC filter to isolate it from the rest of the circuitry. It directly feeds the sensor. The sensor doesn't seem to mind the noisy rail it is now getting. A simple 1st order RC filter on the sensor's output is sufficient to knock the noise down to a level that renders it insignificant.

Long story short - no real issues with noise given the filtering I've implemented.