On Monday, August 10, 2020 at 9:43:06 AM UTC-4, David Brown wrote:
> On 08/08/2020 01:37, Chris wrote:
> > On 08/07/20 14:57, Phil Hobbs wrote:
> > 
> >> I know it isn't what the cool kids do, but that doesn't make it
> >> unreliable or cause it to fail safety inspections AFAICT. Medical may be
> >> different, but it would pass UL with no issues. What regulations nwould
> >> you expect it to fall foul of?
> >>
> >> Cheers
> >>
> >> Phil Hobbs
> >>
> > 
> > It's not the regulations, as afaics approvals are more concerned with
> > *process*, how the design decisons were made and why. Would not
> > necessarily catch design errors that resulted in a regulator running
> > near it's design limit, though avionics or mil temperature cycle
> > testing might  cause a failure because of lack of margin for a device.
> > 
> > Whatever you might think, it's well known that high temperatures and
> > cycling shortens the life of electronics and no, I don't have
> > references,but it's all on the web. Some prefer to skimp every penny,
> > others prefer conservative design rules. It's the difference between
> > consumer electronics and pro kit. may look like the same parts, but
> > the design tradeoffs are different. Engineering design is always an
> > approximation and compromise: performance, cost, reliability, pick any
> > two...
> > 
> > Chris
> > 
> 
> I think it's fair to say that this 70 C case with poor thermal contact 
> with the board is an indicator of bad design - bad schematic design 
> (using a linear regulator rather than a switch mode), and bad layout 
> design (poor cooling).  And poor design in such an obvious case suggests 
> poor design overall - which is not going to go well for any kind of 
> safety requirements or regulations where the design process is considered.

But you're not judging, right? 

No one supports the use of a switching regulator... well, until I did a power budget.  Seems that the two LCD displays use some 10x the power of the rest of the circuit ex motor.  With the motor using 2 amps at some duty cycle the LCD displays appear to be more of a drain on the battery.  We are looking at the issue now.  Even with a switcher the savings is only 50% of the LCD power.  

Personally I think LCDs suck with a contrast ratio of 3:1, but the rest of the world seems to like them. 


> It is also correct to say that the reliability of a system is a function 
> of the reliability of all the parts, and that high temperatures and 
> temperature cycling shorten lifetimes and increase the rate of failures. 
>   But you can't say how significant this particular part would be 
> without a lot more information about the rest of the system.  As Phil 
> says, a part rated for 125 C is not going to see much increased failure 
> risk at 70 C compared to 50 C.  But if it sits next to a part that is 
> rated to 85 C and the final system is mounted in a closed box and used 
> in the Qatar summer, perhaps it will have a relevant effect.
> 
> All in all, this looks like a rush job put together by people with no 
> experience with high reliability or safety-related systems.  That's the 
> general problem, rather than one particular hot regulator.

That's certainly true.  The other day I was in a conference call with the project lead and the new requirements guy.  The requirements are more of a list than anything formally analyzed.  The ultimate reason for this seems to be the realization of the large amounts of personal time being invested in the project and in particular the time invested by the project lead.  

For me the project is recreation during the pandemic isolation.  For others it competes for their free time against families, etc.  That doesn't change the fact that to do the job right requires using the proper tools and methods. 

I am nearing my bail out point.  I'd like to get through a design review of the schematic and the layout.  Then I'm happy to go. 

-- 

  Rick C.

  -++- Get 1,000 miles of free Supercharging
  -++- Tesla referral code - https://ts.la/richard11209

On 08/08/2020 01:37, Chris wrote:
> On 08/07/20 14:57, Phil Hobbs wrote:
> 
>> I know it isn't what the cool kids do, but that doesn't make it
>> unreliable or cause it to fail safety inspections AFAICT. Medical may be
>> different, but it would pass UL with no issues. What regulations nwould
>> you expect it to fall foul of?
>>
>> Cheers
>>
>> Phil Hobbs
>>
> 
> It's not the regulations, as afaics approvals are more concerned with
> *process*, how the design decisons were made and why. Would not
> necessarily catch design errors that resulted in a regulator running
> near it's design limit, though avionics or mil temperature cycle
> testing might  cause a failure because of lack of margin for a device.
> 
> Whatever you might think, it's well known that high temperatures and
> cycling shortens the life of electronics and no, I don't have
> references,but it's all on the web. Some prefer to skimp every penny,
> others prefer conservative design rules. It's the difference between
> consumer electronics and pro kit. may look like the same parts, but
> the design tradeoffs are different. Engineering design is always an
> approximation and compromise: performance, cost, reliability, pick any
> two...
> 
> Chris
> 

I think it's fair to say that this 70 C case with poor thermal contact 
with the board is an indicator of bad design - bad schematic design 
(using a linear regulator rather than a switch mode), and bad layout 
design (poor cooling).  And poor design in such an obvious case suggests 
poor design overall - which is not going to go well for any kind of 
safety requirements or regulations where the design process is considered.

It is also correct to say that the reliability of a system is a function 
of the reliability of all the parts, and that high temperatures and 
temperature cycling shorten lifetimes and increase the rate of failures. 
  But you can't say how significant this particular part would be 
without a lot more information about the rest of the system.  As Phil 
says, a part rated for 125 C is not going to see much increased failure 
risk at 70 C compared to 50 C.  But if it sits next to a part that is 
rated to 85 C and the final system is mounted in a closed box and used 
in the Qatar summer, perhaps it will have a relevant effect.

All in all, this looks like a rush job put together by people with no 
experience with high reliability or safety-related systems.  That's the 
general problem, rather than one particular hot regulator.

On Sat, 08 Aug 2020 00:37:35 +0100, Chris <xxx.syseng.yyy@gfsys.co.uk>
wrote:

>On 08/07/20 14:57, Phil Hobbs wrote:
>
>> I know it isn't what the cool kids do, but that doesn't make it
>> unreliable or cause it to fail safety inspections AFAICT. Medical may be
>> different, but it would pass UL with no issues. What regulations nwould
>> you expect it to fall foul of?
>>
>> Cheers
>>
>> Phil Hobbs
>>
>
>It's not the regulations, as afaics approvals are more concerned with
>*process*, how the design decisons were made and why. Would not
>necessarily catch design errors that resulted in a regulator running
>near it's design limit, though avionics or mil temperature cycle
>testing might  cause a failure because of lack of margin for a device.

At what ambient temperature and air pressure was that +70 C case
temperature taken ? At +20 C and sea level ?

If the product is aimed for the international market, people live in
over +40 C temperatures and some at over 4 km altitude. With
convection cooling, the case (and junction temperature) is
proportional to ambient temperature.

At higher altitudes, the air pressure and density is lower, thus a
higher air flow (fan ?) us required to restore the mass flow and thus
heat removal at a specific delta-T.

On Friday, August 7, 2020 at 9:10:30 PM UTC-4, Paul Rubin wrote:
> Rick C <gnuarm.deletethisbit@gmail.com> writes:
> > BTW, does anyone understand what Phil Hobbs is going on about???  Why
> > does he talk in code?
> 
> I think part of the idea might be that linear regulation is very simple
> compared to switchers, so it has fewer ways to go wrong despite lower
> efficiency.  Either way though, I'm used to regulated P/S having an
> overvoltage crowbar that shuts things down if the regulation fails, and
> in the case of a ventilator it is presumably never ok to shut down.  So:
> backups for the backups.

I don't recall anyone talking about failure other than higher temperatures resulting in higher failure rates.  

I don't know what he was going on about.  The layout of the board was ducked up.  The guy forgot to override the default thermal isolation of the thermal pad on the regulator.  The next board spin won't have that problem and the regulator won't shut down from excessive heat in the regulator.  

No one in the design group has said the device has to have any special reliability rating.  

I remember working on a military radio with secure encryption.  There was an odd circuit with many switches.  I could not find anyone who could explain how it worked, but it seemed to have to do with no single point of failure sort of thing.  When I say "many", I mean MANY, like fifty separate SPST SC-70 devices.  When a circuit has built in redundancy, how do you check to assure the redundancy is working?  I guess they would have to add a means of disabling the redundant part as well as the primary part separately.  But then how do you know your disabling circuit is working???  

At some point there has to be a single point of failure.  In nuclear power plants, if nowhere else, it is the procedures.  If they are wrong, they are wrong through and through with no back up.  The North Anna reactors went offline in the earthquake a few years ago.  One of the diesel generators failed.  When they researched the failure they found it was due to a faulty installation procedure that was common to all of the generators.  A single point of failure. 

-- 

  Rick C.

  -+-+ Get 1,000 miles of free Supercharging
  -+-+ Tesla referral code - https://ts.la/richard11209

Rick C <gnuarm.deletethisbit@gmail.com> writes:
> BTW, does anyone understand what Phil Hobbs is going on about???  Why
> does he talk in code?

I think part of the idea might be that linear regulation is very simple
compared to switchers, so it has fewer ways to go wrong despite lower
efficiency.  Either way though, I'm used to regulated P/S having an
overvoltage crowbar that shuts things down if the regulation fails, and
in the case of a ventilator it is presumably never ok to shut down.  So:
backups for the backups.

On Friday, August 7, 2020 at 8:30:17 PM UTC-4, Phil Hobbs wrote:
> On 2020-08-07 19:00, Chris wrote:
> > On 08/07/20 15:12, Rick C wrote:
> >>
> > 
> >> Efficiency may be personally important to an engineer who wants to 
> >> overdesign everything, but this device has a 2 amp motor running off 
> >> the 15 volts and it is all line powered, so no need to squeeze every 
> >> last joule out of the design.  We are measuring some low level signals 
> >> that could easily be disrupted by the switching currents.  So adding 
> >> that type of a noise source would only be done if actually required.
> >>
> >> Oddly enough most of the power used in the electronics other than the 
> >> motor are in the LED back lights for the low power LCD displays.  I 
> >> have thought of using a switcher for them, but even at the 250 mA 
> >> power level it's just not worth the hassle of dealing with the noise.
> >>
> >> Hmmm...  maybe that is a bigger impact than anyone here is thinking.  
> >> The display is on all the time.  The motor only runs at high power 
> >> about half the time.  So it's really more like 1 amp.  Now the ratio 
> >> is more like 4:1.  We don't have good data on the motor current, so no 
> >> one has paid attention to the battery time.
> >>
> >> There are so many holes in this net.  lol
> >>
> >> BTW, does anyone understand what Phil Hobbs is going on about???  Why 
> >> does he talk in code?
> >>
> > 
> > If you are switching a 2 amp motor load, then you already have more
> > problems than a point of use regulator, in terms of supply load step
> > function. If it's a dc brush motor, perhaps permanent magnet, that
> > looks like little more than 0 ohms, or nearly a short circuit at rest,
> > so you will need to ramp that up to limit current. Once spun up,
> > power is more or less proportional to load, but unless the data
> > sheet has all the info, you will need to build a mechanical test rig
> > to define the motor characteristics. I would do that anyway, to get
> > a idea of what size motor is needed and optimise the circuit
> > to drive it.
> > 
> > There's quite range of ic's for motor control, with soft start,
> > current limiting etc. You could perhaps use a brushless moter and
> > again, some ic choice on how that's done. Advantage is zero
> > commutator noise to upset other things. You really can't just use a
> > relay to switch that motor.
> > 
> >  > BTW, does anyone understand what Phil Hobbs is going on about???
> >    Why does he talk in code?
> > 
> > Iirc, if he will excuse the comment, Phil is a scientist, Science is
> > about math and numbers, measurement accuracy, whereas electronics,
> > once you get the theory  under the belt, as more about experience,
> > knowing what works, what doesn't and not a little intuition...
> > 
> > Chris
> > 
> 
> I've been an electronics designer for a decade longer than I've been a 
> scientist.

And yet you still don't understand anything we are talking about.  

-- 

  Rick C.

  -+-- Get 1,000 miles of free Supercharging
  -+-- Tesla referral code - https://ts.la/richard11209

On Friday, August 7, 2020 at 7:00:47 PM UTC-4, Chris wrote:
> On 08/07/20 15:12, Rick C wrote:
> >
> 
> > Efficiency may be personally important to an engineer who wants to overdesign everything, but this device has a 2 amp motor running off the 15 volts and it is all line powered, so no need to squeeze every last joule out of the design.  We are measuring some low level signals that could easily be disrupted by the switching currents.  So adding that type of a noise source would only be done if actually required.
> >
> > Oddly enough most of the power used in the electronics other than the motor are in the LED back lights for the low power LCD displays.  I have thought of using a switcher for them, but even at the 250 mA power level it's just not worth the hassle of dealing with the noise.
> >
> > Hmmm...  maybe that is a bigger impact than anyone here is thinking.  The display is on all the time.  The motor only runs at high power about half the time.  So it's really more like 1 amp.  Now the ratio is more like 4:1.  We don't have good data on the motor current, so no one has paid attention to the battery time.
> >
> > There are so many holes in this net.  lol
> >
> > BTW, does anyone understand what Phil Hobbs is going on about???  Why does he talk in code?
> >
> 
> If you are switching a 2 amp motor load, then you already have more
> problems than a point of use regulator, in terms of supply load step
> function. If it's a dc brush motor, perhaps permanent magnet, that
> looks like little more than 0 ohms, or nearly a short circuit at rest,
> so you will need to ramp that up to limit current. Once spun up,
> power is more or less proportional to load, but unless the data
> sheet has all the info, you will need to build a mechanical test rig
> to define the motor characteristics. I would do that anyway, to get
> a idea of what size motor is needed and optimise the circuit
> to drive it.
> 
> There's quite range of ic's for motor control, with soft start,
> current limiting etc. You could perhaps use a brushless moter and
> again, some ic choice on how that's done. Advantage is zero
> commutator noise to upset other things. You really can't just use a
> relay to switch that motor.

Done and dusted.  Not sure why you are going off on this tangent. 


>  > BTW, does anyone understand what Phil Hobbs is going on about???
>    Why does he talk in code?
> 
> Iirc, if he will excuse the comment, Phil is a scientist, Science is
> about math and numbers, measurement accuracy, whereas electronics,
> once you get the theory  under the belt, as more about experience,
> knowing what works, what doesn't and not a little intuition...

Ah, so that's why he says things like, "zilch" with no context?  Got it.  

He seems to like playing mind games... mostly with his own mind.  Talking around things without explanations is just screwing with yourself. 

-- 

  Rick C.

  --++ Get 1,000 miles of free Supercharging
  --++ Tesla referral code - https://ts.la/richard11209

On 2020-08-07 19:00, Chris wrote:
> On 08/07/20 15:12, Rick C wrote:
>>
> 
>> Efficiency may be personally important to an engineer who wants to 
>> overdesign everything, but this device has a 2 amp motor running off 
>> the 15 volts and it is all line powered, so no need to squeeze every 
>> last joule out of the design.  We are measuring some low level signals 
>> that could easily be disrupted by the switching currents.  So adding 
>> that type of a noise source would only be done if actually required.
>>
>> Oddly enough most of the power used in the electronics other than the 
>> motor are in the LED back lights for the low power LCD displays.  I 
>> have thought of using a switcher for them, but even at the 250 mA 
>> power level it's just not worth the hassle of dealing with the noise.
>>
>> Hmmm...  maybe that is a bigger impact than anyone here is thinking.  
>> The display is on all the time.  The motor only runs at high power 
>> about half the time.  So it's really more like 1 amp.  Now the ratio 
>> is more like 4:1.  We don't have good data on the motor current, so no 
>> one has paid attention to the battery time.
>>
>> There are so many holes in this net.  lol
>>
>> BTW, does anyone understand what Phil Hobbs is going on about???  Why 
>> does he talk in code?
>>
> 
> If you are switching a 2 amp motor load, then you already have more
> problems than a point of use regulator, in terms of supply load step
> function. If it's a dc brush motor, perhaps permanent magnet, that
> looks like little more than 0 ohms, or nearly a short circuit at rest,
> so you will need to ramp that up to limit current. Once spun up,
> power is more or less proportional to load, but unless the data
> sheet has all the info, you will need to build a mechanical test rig
> to define the motor characteristics. I would do that anyway, to get
> a idea of what size motor is needed and optimise the circuit
> to drive it.
> 
> There's quite range of ic's for motor control, with soft start,
> current limiting etc. You could perhaps use a brushless moter and
> again, some ic choice on how that's done. Advantage is zero
> commutator noise to upset other things. You really can't just use a
> relay to switch that motor.
> 
>  > BTW, does anyone understand what Phil Hobbs is going on about???
>    Why does he talk in code?
> 
> Iirc, if he will excuse the comment, Phil is a scientist, Science is
> about math and numbers, measurement accuracy, whereas electronics,
> once you get the theory  under the belt, as more about experience,
> knowing what works, what doesn't and not a little intuition...
> 
> Chris
> 

I've been an electronics designer for a decade longer than I've been a 
scientist.

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

On 2020-08-07 19:37, Chris wrote:
> On 08/07/20 14:57, Phil Hobbs wrote:
> 
>> I know it isn't what the cool kids do, but that doesn't make it
>> unreliable or cause it to fail safety inspections AFAICT. Medical may be
>> different, but it would pass UL with no issues. What regulations nwould
>> you expect it to fall foul of?
>>
>> Cheers
>>
>> Phil Hobbs
>>
> 
> It's not the regulations, as afaics approvals are more concerned with
> *process*, how the design decisons were made and why. Would not
> necessarily catch design errors that resulted in a regulator running
> near it's design limit, though avionics or mil temperature cycle
> testing might  cause a failure because of lack of margin for a device.
> 
> Whatever you might think, it's well known that high temperatures and
> cycling shortens the life of electronics and no, I don't have
> references,but it's all on the web. Some prefer to skimp every penny,
> others prefer conservative design rules. It's the difference between
> consumer electronics and pro kit. may look like the same parts, but
> the design tradeoffs are different. Engineering design is always an
> approximation and compromise: performance, cost, reliability, pick any
> two...
> 
> Chris

Okay, so you don't have anything either.  No worries, thanks.

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

On 08/07/20 14:57, Phil Hobbs wrote:

> I know it isn't what the cool kids do, but that doesn't make it
> unreliable or cause it to fail safety inspections AFAICT. Medical may be
> different, but it would pass UL with no issues. What regulations nwould
> you expect it to fall foul of?
>
> Cheers
>
> Phil Hobbs
>

It's not the regulations, as afaics approvals are more concerned with
*process*, how the design decisons were made and why. Would not
necessarily catch design errors that resulted in a regulator running
near it's design limit, though avionics or mil temperature cycle
testing might  cause a failure because of lack of margin for a device.

Whatever you might think, it's well known that high temperatures and
cycling shortens the life of electronics and no, I don't have
references,but it's all on the web. Some prefer to skimp every penny,
others prefer conservative design rules. It's the difference between
consumer electronics and pro kit. may look like the same parts, but
the design tradeoffs are different. Engineering design is always an
approximation and compromise: performance, cost, reliability, pick any
two...

Chris