I'm designing an 8 channel, high performance thermocouple input module for industrial applications. The specification requires full safety isolation between the channels (250VAC). The problem is using a separate dc-dc converter transformer for each channel takes too much board space. Does anyone have any references or recommendations on the design of a single transformer with 6-8 fully isolated secondaries? The dc-dc input in 24VDC and the output requirement is +5V per channel. Power rating is on the order of 2-3 watts.
Low power dc-dc converter transformer with multiple isolated secondaries
Dec 05, 2006
14 Replies
Since you are talking thermocouples you must also be talking low speed. Perhaps use a capacitor coupled bridge to transfer power using a CMOS push-pull stage (that is two Y2 caps to get safety isolation and a bridge to rectify the pulses). Then communicate back to the primary side using a clamp to short the bridge. Measure current consumption on the primary side to get the info back. This is low speed, but it should be do-able for your application
Costs = nothing, but R&D time is larger than a brute force transformer solution
Regards
Klaus
Often I have to design such features into medical gear. There we need up to 5KV of isolation but the method always boils down to this:
#77 or similar ferrite core. My core preference is the toroid but that can get old if you need lots of windings. One primary winding that is driven by an H-bridge plus some EMI muffling. This primary is made of wire with a 100% certified insulation. The drive is somewhere between
40kHz and 200kHz, usually. For lower power apps I have used cheap audio amps as a driver.Then as many secondaries as desired. If they need high isolation against each other you also need to use the good stuff for those windings. That is rectified and regulated either by a linear circuit or another switcher. Where every penny counts I often use two BAV99 in a bridge configuration.
Regards, Joerg
http://www.analogconsultants.com
formatting link
We just used a transformer per channel. Little bitty ones.
How big does your board have to be?
John
What kind of certification does engineering medical equipment require..? (and where to get 'life support' chips?)
Usually EN601, the whole enchilada. This also includes EMC, of course. In my own designs I go a step further in that I insist that they be defibrillator proof even if the regs don't require it.
You almost have to make your own. Seriously, that's what one company I know does. I am not aware of any large manufacturer that doesn't have a life support exclusion. Of course, you can always negotiate but then there needs to be a huge stash of money on the table or they won't even talk to you about it.
Regards, Joerg
http://www.analogconsultants.com
Was that the telco transformer from the other thread? How do you manage the 15KV ESD?
Regards, Joerg
http://www.analogconsultants.com
Yup, the $1 ISDN things, with 2N7002 drivers. They're rated for the
1500 volt telecom transient. Some modest capacitors between the input pins and ground (good for emi anyhow) divide down the human-body-model zot to below 1500.John
Aha. Caps. That's what I am usually not allowed to do on my medical designs. They have to stomach the discharge of a 32uF cap at 5000V into a body model via a huge inductor. When the agency folks want me to demonstrate that their engineers move to the far corner of the room before I push that big red button.
Regards, Joerg
http://www.analogconsultants.com
Is this any good to you? (Not 2-3 Watts, but maybe that isn't needed)
formatting link
formatting link
Chris
Is it a written exam or something else?
I guess 'defibrillator proof' means you can put the pads on your equipment and it will still work.
What about haveing an extra quality ensurance, no single point of failure plus an insurance for any legal claims?
Makeing your own chips doesn't seem trivial considering todays complexity .. ;)
What other thread?
James T. White
Nope, its a set of standards by which we must live.
There are several grades in terms of how the equipment is allowed react, depending on its function during a procedure. All the way from "must not electrically break down" to "must be fully operational again in x seconds". I always make sure it's operational in under a second no matter what. Never failed one test :-)
That's the negotiating part. However, large manufacturers won't spend that time with you unless there is a substantial business volume in the cards for them.
But it's done. Out of sheer necessity. The good thing is that there are a lot more independent foundries than there were 20 years ago.
Regards, Joerg
http://www.analogconsultants.com
Check John Larkin's June 17, 2005 posts in this thread:
formatting link
I believe there was another more recent disussion though.
Regards, Joerg
http://www.analogconsultants.com
Severall companies make multi-output transformers. This one has six windings:
formatting link
Maybe they have an 8-output one.
I often designed my own using multisection bobbins. One time I used a toroid with wire loops from each section. That was a "driven" transformer design. I used 1000V wire to get low capacitance and high breakdown. Drive the primary push-pull from your 24v and all you'll need is a single turn on the secondary.
Works.
Steve.
Steven D. Swift, novatech@eskimo.com, http://www.novatech-instr.com
NOVATECH INSTRUMENTS, INC. P.O. Box 55997
206.301.8986, fax 206.363.4367 Seattle, Washington 98155 USA
Join the Discussion
Have something to add? Share your thoughts — no account required.
Didn't find your answer?
Ask the community — no account required