Ramblings about Airborne Power supply design and testing

As I stated, I should have made item 16 a separate post, it really was not the meat of what i was trying to convey

I have been frequently assigned to power supply testing (and some design) per MIL-STD-704 and DO-160.

normal operation" - which have a few subtleties) the important thing is tha t if 28V (typical nominal voltage for my category of equipment) is present on the input the box better be working. It is surprising how difficult tha t simple requirement is to achieve. The testing regiment requires the unit to be put through many types of over/under voltage tests that will take th e box down and then the requirement is that the unit re-powers up without u ser intervention and works correctly. This is not a trivial requirement. Much of the testing really revolves around the standard's notion that this is not a trivial requirement.

e and beyond those in DO-160. In one case they require 8 waveforms that im pose glitches on the startup voltage when powering up. Sure enough, we ran these waveforms and one of their sequences locked our unit up (power was a pplied to the terminals and the unit was latched up and required user inter vention to power it off and back on to get it to work).

.....oh boy , this requires a lot of attention to get it right.

t times, to educate my management that most of the time I find a problem it is in the setting up of the required waveforms. Say I have an undervoltag e requirement of 12 Volts and as I am setting it up I apply 11 volts and th e unit locks up. The first thing a manager tries to do (until properly edu cated....fortunately my mangers do try to do the right thing....most of the time) is to ask if the unit works with the specified waveform and I have t o answer yes.....but.....The real purpose of the test is to make you look a t the unit 100 different ways to Sunday to see if the unit latches up and p art of the process is the setup variations that will frequently be where th e sweet (sour) spot is. So I have to go back and remind them that if I can have 28V appllied to the box and it is latched up it really does not matter what I did to the unit before ---the box not adequate for an aircraft. (o f course this whatever you do are within reasonable guidelines....undervolt ages, short duration overvoltages and some ripple on the input voltage). W hen a manager tries to tell me to just run the waveforms I ask him if we ca n have this discussion with the customer to get clarification.....99% of th e time that forces the required (and painful) design changes.

k MIL-STD-704/DO-160 is rigorous you should see what one well known aircraf t manufacturer in France requires in addition to DO-160. It takes weeks to run the tests.....They do not want a box to latch up on their aircraft

pplied that the unit is expected to work through. The aircraft manufacture rs HATE HATE HATE latchups, but they also do not like the unit temporarily shutting off when it should not shut off. These normal operation requiremen ts revolve around3 basic things 1. the unit will stay powered up for 200 ms (this can vary) when power drops below a specified voltage (hold-up cap re quired) 2. The unit will stay powered up when over voltages are less than 2

00 ms 3. The unit will operate through various audio frequencies imposed on top of the DC voltage. 4. The unit will operate from typically 16or18V to approximately 29-33 volts (each unit will have these nailed down exactly in a spec)

ot cause the unit to latch up (per above) and also not damage the unit (OK thats obvious, but it is explicitly stated and required).

n Ideal diode (smart FET) is used instead of a schottky diode on the input. Yeah they work for negative voltages....but About those Audio frequencie s imposed on the DC lines..... , they walk right through the ideal diode. I am of the persuasion to use a schottky diode with a large capacitor afte r it. This will not allow internal resonances to develop which can cause m any amps of current to circulate through the input circuitry. Take the 0.5 V hit on efficiency. The real diode is so much nicer (do note I am coming at this for units that are 30W or less, usually less).

lications of a particular interrupt (for instance) waveform. I insist on d oing each one manually and slowly. Apply and observe before the next appli cation of the transient. I have seen where people try to automate these tr ansient applications and wind up making a burst of 5. I think this breaks the spirit of what is expected. I have seen the automatic test blow through them so fast that you cannot even tell what happened. I insist on manual application of transient waveforms.

well of audio frequncy ripple is OK if you are still able to properly monit or the equipment.

ests are not about how the unit operates with voltage corruptions, but rath er it is an assessment of how your unit might affect the power bus. For lo w power equipment always fight for these things to be characterized but not specified. EVERY TIME I test any of these items to a spec it fails....Esp ecially inrush current. Just characterize it!

a hard spec. It will fail. Fight to characterize it. Low power stuff is not going to take down the aircraft power bus and your internal capacitors have to get charged up and yes, your EMI caps theoretically take infinite current for short durations to charge up. When you design around inrush yo u start to take away internal capacitance that you really want. The aircra ft has to tranfer energy into your unit on power up. I think the aircraft manufacturers get that but the designers see a spec as a challenge and try to meet it (at the expense of inadequate internal capacitance or being in t he crappy position of telling your management that you failed at test time) ...Actually, the customer will probably give you relief on it (not on anyth ing else though)so fight up front to avoid the headaches.

Item 17: The link here :

shows the picker gradient amplifier that I use to generate all kind of MIL- STD-704 compliant waveforms. These things are amazing! +/- 100 ish V @ 20 amp output and 200Khz BW. They are indestructible. you can drive this wit h a 2K AWG. If you have a calibrated scope and calibrated power meter (thr ow in required transformer for another 1K) you can be doing a lot of these tests for credit instead of going to a 3-5K/day test lab. And you can do a ll your own dry runs. You can also show your power supply designer where a dditional work is required :)

nd I have been frequently assigned to power supply testing (and some design ) per MIL-STD-704 and DO-160.

("normal operation" - which have a few subtleties) the important thing is t hat if 28V (typical nominal voltage for my category of equipment) is presen t on the input the box better be working. It is surprising how difficult t hat simple requirement is to achieve. The testing regiment requires the un it to be put through many types of over/under voltage tests that will take the box down and then the requirement is that the unit re-powers up without user intervention and works correctly. This is not a trivial requirement. Much of the testing really revolves around the standard's notion that thi s is not a trivial requirement.

ove and beyond those in DO-160. In one case they require 8 waveforms that impose glitches on the startup voltage when powering up. Sure enough, we r an these waveforms and one of their sequences locked our unit up (power was applied to the terminals and the unit was latched up and required user int ervention to power it off and back on to get it to work).

wn.....oh boy , this requires a lot of attention to get it right.

at times, to educate my management that most of the time I find a problem it is in the setting up of the required waveforms. Say I have an undervolt age requirement of 12 Volts and as I am setting it up I apply 11 volts and the unit locks up. The first thing a manager tries to do (until properly e ducated....fortunately my mangers do try to do the right thing....most of t he time) is to ask if the unit works with the specified waveform and I have to answer yes.....but.....The real purpose of the test is to make you look at the unit 100 different ways to Sunday to see if the unit latches up and part of the process is the setup variations that will frequently be where the sweet (sour) spot is. So I have to go back and remind them that if I ca n have 28V appllied to the box and it is latched up it really does not matt er what I did to the unit before ---the box not adequate for an aircraft. (of course this whatever you do are within reasonable guidelines....undervo ltages, short duration overvoltages and some ripple on the input voltage). When a manager tries to tell me to just run the waveforms I ask him if we can have this discussion with the customer to get clarification.....99% of the time that forces the required (and painful) design changes.

ink MIL-STD-704/DO-160 is rigorous you should see what one well known aircr aft manufacturer in France requires in addition to DO-160. It takes weeks to run the tests.....They do not want a box to latch up on their aircraft

applied that the unit is expected to work through. The aircraft manufactu rers HATE HATE HATE latchups, but they also do not like the unit temporaril y shutting off when it should not shut off. These normal operation requirem ents revolve around3 basic things 1. the unit will stay powered up for 200 ms (this can vary) when power drops below a specified voltage (hold-up cap required) 2. The unit will stay powered up when over voltages are less than 200 ms 3. The unit will operate through various audio frequencies imposed on top of the DC voltage. 4. The unit will operate from typically 16or18V t o approximately 29-33 volts (each unit will have these nailed down exactly in a spec)

not cause the unit to latch up (per above) and also not damage the unit (O K thats obvious, but it is explicitly stated and required).

an Ideal diode (smart FET) is used instead of a schottky diode on the inpu t. Yeah they work for negative voltages....but About those Audio frequenc ies imposed on the DC lines..... , they walk right through the ideal diode . I am of the persuasion to use a schottky diode with a large capacitor af ter it. This will not allow internal resonances to develop which can cause many amps of current to circulate through the input circuitry. Take the 0 .5V hit on efficiency. The real diode is so much nicer (do note I am comin g at this for units that are 30W or less, usually less).

pplications of a particular interrupt (for instance) waveform. I insist on doing each one manually and slowly. Apply and observe before the next app lication of the transient. I have seen where people try to automate these transient applications and wind up making a burst of 5. I think this break s the spirit of what is expected. I have seen the automatic test blow throu gh them so fast that you cannot even tell what happened. I insist on manua l application of transient waveforms.

dwell of audio frequncy ripple is OK if you are still able to properly mon itor the equipment.

tests are not about how the unit operates with voltage corruptions, but ra ther it is an assessment of how your unit might affect the power bus. For low power equipment always fight for these things to be characterized but n ot specified. EVERY TIME I test any of these items to a spec it fails....E specially inrush current. Just characterize it!

st a hard spec. It will fail. Fight to characterize it. Low power stuff is not going to take down the aircraft power bus and your internal capacito rs have to get charged up and yes, your EMI caps theoretically take infinit e current for short durations to charge up. When you design around inrush you start to take away internal capacitance that you really want. The airc raft has to tranfer energy into your unit on power up. I think the aircraf t manufacturers get that but the designers see a spec as a challenge and tr y to meet it (at the expense of inadequate internal capacitance or being in the crappy position of telling your management that you failed at test tim e)...Actually, the customer will probably give you relief on it (not on any thing else though)so fight up front to avoid the headaches.

L-STD-704 compliant waveforms. These things are amazing! +/- 100 ish V @ 2

0 amp output and 200Khz BW. They are indestructible. you can drive this w ith a 2K AWG. If you have a calibrated scope and calibrated power meter (t hrow in required transformer for another 1K) you can be doing a lot of thes e tests for credit instead of going to a 3-5K/day test lab. And you can do all your own dry runs. You can also show your power supply designer where additional work is required :)

no... this is not my stuff for sale and this has not been a lead in to sell these items. But I do make a great recommendation for these bad boys :)

TL;DR

If you broke that up into bite-size chunks, I think you'd get a much better response.

Cheers

Phil Hobbs

-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510

TL;DR

You and 7 billion other people on planet earth did no read it either. Thanks for weighing in though

You're entirely welcome. Unlike most of them (and you), I did offer a helpful suggestion.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

So telling someone that TL-DR is helpful? Normally when i wish to help some one I read the content and then comment on the content. If I am immediately turned off I may be rude and tell them so, but I don't think I would ever think in my mind i did them a favor. Anyhow, I wrote this because i had an hour and wanted to add content to the forum about a topic i have something to contribute. Most of my design stuff i can't really discuss. Also i did i t for myself because I may wish to recall these thoughts at some later time and now i can

So are you the OP or his mum? Otherwise, where do you get off, lecturing folks like that?

Since you apparently didn't get past the first line of my first post, allow me to spell it out.

The helpful bit was "If you broke that up into bite-size chunks, I think you'd get a much better response."

Which I think is entirely true and reasonable. If you don't, oh, well.

Cheers

Phil Hobbs

You have a very high opinion of yourself. I mean you seriously think telli ng someone you looked at what they wrote and immediately dismissed it becau se it was too long is offering help. Maybe I do not care,if PH disses my p ost. I do know tjat PH did not have the time to read it but he sure had the time to post about how he immediately dismissed it. I mean that is first r ate help. How much do i owe you for this help?

Ah, so you're the OP via a sock puppet, got it. Have a nice day.

Cheers

Phil Hobbs

snipped-for-privacy@columbus.rr.com wrote in news:f4eedbfd-abea-4355-8303- snipped-for-privacy@googlegroups.com:

I think computer science 'kids' should have to start out with the altair and then the XT, and upright video game processing and function.

They are ALL jaded and want the latest and greatest hot 3D video game capable screamer. They learn nothing of the nuts and bolts.

Maybe put them on ARCNET network, and slowly wean them up on faster and faster hooks, all the while explaining the difference between a

22kB executable file (say NetHack), and a 4GB movie file. Until they see how we had it. What was a good number over the modem? About 15 minutes per Megabyte. Once megabyte files were common.

My first 16MB of RAM was almost $600 used.

My first ONE GigaByte hard drive was half the size of a shoebox and cost me $600 used. (OMG I feel old all of a sudden).

It shows automotive hood springs. Recheck the number.

You might do a pre-qual assurance, but the calibrations and ref tracing required for most certifications are outside the budget of the average design/test facility in power conversion field. This requires deep pockets.

You might be disappointed at the 'expertise' offered at a competative test facility. Recommend your product guru be there to hold hands to avoid most issues.

For extraneous specs, in-house hardware capability will always be useful, but likely with equipment bought surplus or on e-bay, as you suggest.

RL

Used to be, a good power supply kept working, in flames if necessary, until the input source was removed.

Second-guessing the operator was considered bad form.

RL

snipped-for-privacy@columbus.rr.com wrote in news:19082fcf-4f5e-4014-be3e- snipped-for-privacy@googlegroups.com:

When I did HV power supply design for high altitude operation, which was set to 37k ft for passenger craft and we had to do 70,000 ft spec for military gear.

So, we did a triple output anode supply for the HUGE 30" x 42" x

8" optical projector unit that was used on very old 747s way back when a projector was the state of the art.

We also refitted and Tempest shielded (bottled that sucker up tight)a 25 inch CRT display to make it into a high altitude aircraft rated device that would not burn up on them. It was a canned display inside a can for rack mounting. Way back in the big, bubble face days too. Pre-vertical flat or anything. Behemoth!

The main consideration for us was thermal management in the thinner cabin air. The math on the latent heat capacity of the air mass had to be considered in cases where down here, the discreet device would radiate sufficiently on its own, might need a finned sink and a forced air flow placed onto it to suffice at the cabin pressure at

70,000 ft. Typically heated air too (of course). Dang! Double difficult.

My latest hire, we sent directly to the test department for a few months. Engineers need to design stuff that is easy to test and cal. And they need to design test sets. He understands that now.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

That is bizarre. I continue to have the picker gradient amplifiers displayed.

For power input testing I have been able to do "for credit" testing because the only items that need to be in calibration is an o-scope and voltage meters....And a current probe. But for many other tests outside labs are needed.

I am trying to see if I can get my picker to run the AC waveforms and do for credit testing on those items too. ( I am running low power equip and that certainly makes doing it yourself easier)

Sounds like a project that was fun to be on.

gainst a hard spec. It will fail. Fight to characterize it. Low power st uff is not going to take down the aircraft power bus and your internal capa citors have to get charged up and yes, your EMI caps theoretically take inf inite current for short durations to charge up. When you design around inr ush you start to take away internal capacitance that you really want. The aircraft has to tranfer energy into your unit on power up. I think the air craft manufacturers get that but the designers see a spec as a challenge an d try to meet it (at the expense of inadequate internal capacitance or bein g in the crappy position of telling your management that you failed at test time)...Actually, the customer will probably give you relief on it (not on anything else though)so fight up front to avoid the headaches.

on this....Engineers (young ones rightfully so, I think) hate writing and t esting requires a lot (A LOT) of thought to what is going to be tested and how to go about doing it. Younger engineers I think should be allowed to l earn the trade and not get bogged down by writing. But, sooner or later, I do not believe you are a good engineer if you cannot write a good test pro cedure. A good procedure is based upon the right data being taken and the right setups. Engineers are normally not good at writing and unfortunately in many corporations the procedure writer is the "lesser" engineer, but th at is a bad wrap ( I have been on both sides ). If you can get past the le sser engineer thing you can learn to use your willingness to write procedur es to your (and the programs ) advantage. The thing about procedures, is t hat nobody wants to look at them until the end of the program, so if you ar e good, you can write the

any respects, writing the tests allows a control over the design the is ver y powerful and goes unchallenged because a design engineer will design to t he test if you can get it through quick enough. I have learned (I am late in my career so I am taking what gets me over the finish line) to embrace w riting procedures and reports and to use them to my (and the programs ) adv antage

ng

f

lds.

t e

their career.

Please keep posting. I have more to learn from you than vice versa.

Perhaps item# 153158735347.

VMEbus?

I had a qual house failing preproduction systems for hipot after humidity soak. Size and cost of small car. They'd screwed up connector plugs for the test, even though we'd provided them with the correct prewired and pretested test harnessing. Change the test connector - no problem.

RL