Why Should I Isolate My Power Supply?

Thread Starter

JackieTee

Joined May 28, 2019
11
What started as a task to reduce noise in a current power supply design has evolved to a complete overhaul of our power supply. Being an intern, I have found myself with a lot of very basic questions.

First one being, how do I know if I need isolation in my design? The power supply takes 200V DC and steps it down to 20V DC to supply 2A to our load. The load is being sent down oil/gas pipeline to detect issues in corrosion, deformation, eccentricity, and so on. Additionally, the tool must operate reliably at temperatures as high as 200C.

I have been doing a lot of research on SMPS and I have struggled immensely with the flyback topology. But, it looks like I might have a viable buck converter design. Can a buck be used for this application? Or do I need the isolation that a flyback provides?

Thanks in advance, guys!
 

ericgibbs

Joined Jan 29, 2010
19,104
hi Jackie,
Welcome to AAC.
I would recommend a transformer isolated SMPS.

MOD:
Posts discussing mains powered, non isolated power supplies are not allowed.

E


 

MaxHeadRoom

Joined Jul 18, 2013
28,827
For this type of application I would look at a linear supply, with this type of installation a SMPS may be prone to electrical storm damage etc, also a consideration is whether you need to earth ground/reference the LV supply.
Max.
 
Last edited:

Thread Starter

JackieTee

Joined May 28, 2019
11
hi Jackie,
Welcome to AAC.
I would recommend a transformer isolated SMPS.

MOD:
Posts discussing mains powered, non isolated power supplies are not allowed.

E

Can you clarify what you mean? Are you saying that you recommend a transformer-isolated SMPS strictly because discussing a non-isolated SMPS is against the rules of the website?

And, if so, is it against the rules to discuss on this site strictly for safety issues? Because, if so, that would be reason enough to isolate my power supply. If it is considered too dangerous to simply discuss the pros/cons of non-isolated topologies, then it is probably too dangerous to send down a natural gas pipeline with all that nice, flammable gas.
 

MaxHeadRoom

Joined Jul 18, 2013
28,827
Is the 200vDC isolated?
If so, there is no need to re-isolate it.
What is the load/current of the LV circuit?
I had a very similar application and successfully used a Zener PS off of 260vdc, the LV demand was a few hundred Ma's.
Max.
 

ericgibbs

Joined Jan 29, 2010
19,104
hi,
What I am saying if you continue to discuss a non isolated supply, the Thread will be closed.
I suggested a transformer isolated supply so that your Thread could stay open, so others can help you.
E
 
hi,
What I am saying if you continue to discuss a non isolated supply, the Thread will be closed.
I suggested a transformer isolated supply so that your Thread could stay open, so others can help you.
E
You should look at some of the other threads, several have discussions on line connected devices and several of these have very dangerous suggestions. I was unaware of this rule, otherwise I would have reported those threads.
 

ericgibbs

Joined Jan 29, 2010
19,104
Hi Charles,
I am concerned when a TS, with apparent limited electrical knowledge, starts asking questions about high voltage devices.

E
 
Sounds like your sending a robot down the pipeline and you need the High voltage to over come the drops?

Typically, say for the 200 V supply, you would have a couple of diode drops above ground. That does not sound like a good idea.

You might have something different going on for the robot.

If it's only 20V 2A, then a simple Linear Transformer psu will be fine,
a Smps will be prone to lighting or other electric pulses, and could breakdown easily.
Max distance was never specified. Pipelines are LONG.

I don't know if we are talking about the 200V or the 20 V supply?
 

crutschow

Joined Mar 14, 2008
34,842
The TS stated he is taking 200Vdc down to 20Vdc.
So it's likely that the 200V is already isolated, in which case additional isolation should not be needed.
 

Thread Starter

JackieTee

Joined May 28, 2019
11
Hi Charles,
I am concerned when a TS, with apparent limited electrical knowledge, starts asking questions about high voltage devices.

E
Don't worry, Eric. While I AM asking about very dangerous subjects and I DO have limited knowledge, I am still just an intern. A huge part of my job is researching, followed by checking with my superiors, and then doing more research. I absolutely NEVER act on my own, I will never be the person to finalize a design, and absolutely no one at my job trusts me enough to just take what I say at face value without double checking my research (for very good and obvious reason). I guarantee that everything discussed in this thread will remain hypothetical unless I happen to stumble upon a solution that the electrical designers, technicians, and project manager pass off on. I repeat - I am merely a student and an intern and NONE of what is discussed here will get implemented or tested in the real world. I am here to learn. Nothing more. If I ask a question about something that is dangerous, it would do me more use to have someone explain to me that it is dangerous and perhaps even go into a bit a detail about why it is dangerous than ban the thread. Teach me! Don't ban me!

Thank you!
 
Hi Charles,
I am concerned when a TS, with apparent limited electrical knowledge, starts asking questions about high voltage devices.

E
I agree completely, they don't understand the dangers involved. I have designed mains connected power supplies in a professional capacity and that stuff scares me. Being afraid of it is a good thing, I don't ever want to become complacent.
 

Thread Starter

JackieTee

Joined May 28, 2019
11
The TS stated he is taking 200Vdc down to 20Vdc.
So it's likely that the 200V is already isolated, in which case additional isolation should not be needed.
I found it out! The 200Vdc supply DOES need to be isolated from the 20Vdc for this particular scenario. While awaiting responses, I ended up getting my answer to my question - which I apologize for my question being so vague in the first place (did I mention that I am new to all of this?). For any future travelers who ends up here with a similar question, I would like to post my findings.

My confusion ultimately began with suddenly being asked to reproduce a buck converter from a datasheet (attached - I ended up replicating Figure 7 on page 23 only I changed the 220m resistor to 33m so I could get my desired output voltage. All of this has been modeled in LTSpice and has yet to be built) that could handle an input of 200V and an output of 20V. This was asked of me after spending months on researching flyback topologies. I got confused as to why my boss would suddenly ask for the transformer isolation to be removed and figured that maybe I misunderstood the concept of isolation on a fundamental level, hence why I came here in the first place.

Upon asking why I was suddenly being asked to design a buck-converter topology I was given the following explanation;

First: This topology will NOT be implemented in our power supply - we WILL be using a flyback topology (we do want the isolation in our final design). I was asked to do the buck converter simply because we wanted to test the LT8316 (this IC is in the attached datasheet). We want to see what sort of noise we get on its output while operating in a manufacturer-designed configuration, and (more importantly) we want to see if it is capable of maintaining that same level of operation at temperatures up to 200C. No company ever tests their parts to that temperature (or at least it is extremely rare) and so we always need to test them ourselves.

Second: We do want to isolate the 200V input from the 20V output in this supply. The explanation I was given for this was actually pretty simple. Basically, we wanted to isolate the 200V side of the power supply from the 20V side to make sure that, if the switch being used in our power supply failed, the 200 volts wouldn't be allowed to pass directly through our supply and destroy our (very expensive) tool.

Thanks everyone! If you happen to run into me again on this site, please bear with me! Until I get a better idea on how to better make use of forums as a resource, I will probably be "mis-asking" a lot of questions!
 

Attachments

ericgibbs

Joined Jan 29, 2010
19,104
hi Jackie,
Thanks for the thoughtful reply, lets see how the Thread goes and hopefully you can gain knowledge from the posted replies.

Eric
 
I ended up replicating Figure 7 on page 23 only I changed the 220m resistor to 33m so I could get my desired output voltage.
The 220mΩ resistor is to set the peak current for the current loop. Reducing it to 33mΩ will allow you to achieve the higher current requirement. The 21.5K and 2.49K resistors on the left side set the output voltage.
 

BobTPH

Joined Jun 5, 2013
9,271
There is a lesson in this thread. If you are a junior person and you don’t understand something about what you are asked to do, ask the person who gave you the assignment!

As someone who was often a project lead, I would much prefer that to having them make an incorrect assumption or wasting time researching a question I could answer immediately,

Bob
 

DickCappels

Joined Aug 21, 2008
10,246
Am I too late to chime in with the following? If so, please ignore.

For conversion from voltages like 200 VDC to 20 VDC it may not be possible to beat the ease of design and the performance of a buck converter. At the currents mentioned (2 A for load) you might get away by direct coupling the base drive to the switch, which would probably be a PNP bipolar.

A long time ago I put a very similar converter into production. The efficiency was very high and because nothing got hot reliability in the field was very good.

The flyback topology is good for low power applications, especially where isolation between the input and output is required, but tends to be less efficient that relatively simple non-isolated buck.
 

Thread Starter

JackieTee

Joined May 28, 2019
11
There is a lesson in this thread. If you are a junior person and you don’t understand something about what you are asked to do, ask the person who gave you the assignment!

As someone who was often a project lead, I would much prefer that to having them make an incorrect assumption or wasting time researching a question I could answer immediately,

Bob
Too be perfectly honest, Bob. I think they brought me on as their intern too early into my college career. Which is very embarrassing to say. The person who hired me seemed to have it in their mind that I would be doing basic tests to check functionality of tools before sending them to customers - something I could handle. They were not expecting me (and neither was I) to be handed assignments to try and reduce noise in and/or help redesign power supplies (among other difficult projects). I get scared to ask questions sometimes. But only because I have to ask so many questions. I worry they will wise up to the fact that I don't know as much as I need to and I will lose the chance to learn. In thirty days at this internship, I have easily learned three times more than everything I've learned thus far from university. If I can keep studying and working hard to make myself useful, I can't IMAGINE where I will be by the end of this year.

Am I too late to chime in with the following? If so, please ignore.

For conversion from voltages like 200 VDC to 20 VDC it may not be possible to beat the ease of design and the performance of a buck converter. At the currents mentioned (2 A for load) you might get away by direct coupling the base drive to the switch, which would probably be a PNP bipolar.

A long time ago I put a very similar converter into production. The efficiency was very high and because nothing got hot reliability in the field was very good.

The flyback topology is good for low power applications, especially where isolation between the input and output is required, but tends to be less efficient that relatively simple non-isolated buck.
The buck converter seems like the easiest topology for us to use, but it looks like the higher-ups really want us to electrically isolate the 200V input from the 20V output. The tools that this PS powers are very expensive for us to make. If something happens, we don't want the 200V to burn through our tool. The main problem is that our current PS is not only operating at only about 60% efficiency, but we get HORRIBLE ringing on the switch which has caused a significant amount of noise/ripple on the output. We're going to be trying out a new Flyback Controller (PWM) which has some built in features to help eliminate ringing in power supplies. I have a simulation in LTSpice which shows a very clean 20V output with virtually no ringing on the switch. We're going to be doing an evaluation board and seeing how it withstands 175C.
 

AnalogKid

Joined Aug 1, 2013
11,200
40 W is not a large load. I would research an 80-100 W off-the-shelf universal input switcher. Many of these can take a DC input directly.

ak
 
Top