OT: Power BJT Power Limit...

[server]

On Wed, 29 Jul 2020 03:38:09 -0700 (PDT), jrwalliker@gmail.com wrote:

On Tuesday, 28 July 2020 17:49:52 UTC+1, whit3rd wrote:
On Tuesday, July 28, 2020 at 4:17:51 AM UTC-7, Phil Allison wrote:
Martin Brown wrote:
--------------------


I recall an infamous PWM X10 power amplifier designed in 1964 by
idiosyncratic but brilliant Ivor Catt

The thin bond out wires eventually failed due to vibration from the high
currents interacting with the Earth\'s magnetic field.

** Not credible.

Well, maybe it is. The technology of the day was weak on switchmode
circuitry, and I\'ve seen exploded-wire failures (in metal cases, just file
off the lid and aim a microscope at the problem).

The dI/dt limit on a few millimeters of wire was due to magnetic
forces, skin effect, and shockwve (ultrasonic) stresses. Not the
terrestrial field, the current-induced field. Unlike thermal failure,
there\'s no blob of melt at the end of the exploded wire... it\'s just
GONE.

I recently tested some pcb mounted fuses to see how well they would
survive the startup surge of a dc input power supply where a large
capacitor was charged up very fast.
After around 100,000 startups the fuse would fail from metal fatigue
at the points where the fuse wire is welded to the connection wires.
The peak current was around 30 to 50 times the rated current of the fuse
for a few tens of microseconds.
I also repeated the test with a powerful magnet on top of the fuse
to see whether this made any difference. It did not.

I think the fuse wire \"instantaneously\" lengthened, thereby changing
the angle of the wire at the terminations, effectively bending it
backwards and forwards slightly each time. The geometry is probably
very similar to that of bond wires in a power transistor.

John

We don\'t use surface-mount fuses. We tested some that would fail at
currents within their hold ratings. They depend too much on PCB
thermals and maybe on luck.

Ditto surface-mount polyfuses. We use leaded ones successfully.

\"Silicon fuses\", \"E-fuses\", have their own problems.

In general, protecting gear from wall-warts is tricky.



--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[server]

On Wednesday, 29 July 2020 15:44:41 UTC+1, jla...@highlandsniptechnology.com wrote:

We don\'t use surface-mount fuses. We tested some that would fail at
currents within their hold ratings. They depend too much on PCB
thermals and maybe on luck.

The fuses I tested were wire leaded, not surface mount.
Surface mount had already been rejected for that project.

John

 

[Kevin Aylward]

\"Phil Allison\" wrote in message
news:27fad0db-46bb-4339-bb48-23c11f1a413eo@googlegroups.com...

Martin Brown wrote:
--------------------

** Has was also know for Wireless World articles about capacitors and
transmission lines being the same.

Until \" Ouida Dogg\" came along and took a few bites out of him.

( Get it \" I\'ve a cat / \" we\'d a dog \" ? )

Hysterical stuff, for conservative WW mag.

Indeed... I got into a tad of a barney with the dude in WW

http://www.ivorcatt.co.uk/x29t.htm


-- Kevin Aylward
http://www.anasoft.co.uk - SuperSpice
http://www.kevinaylward.co.uk/ee/index.html

 

[Phil Hobbs]

On 2020-07-29 06:38, jrwalliker@gmail.com wrote:

On Tuesday, 28 July 2020 17:49:52 UTC+1, whit3rd wrote:
On Tuesday, July 28, 2020 at 4:17:51 AM UTC-7, Phil Allison wrote:
Martin Brown wrote:
--------------------


I recall an infamous PWM X10 power amplifier designed in 1964 by
idiosyncratic but brilliant Ivor Catt

The thin bond out wires eventually failed due to vibration from the high
currents interacting with the Earth\'s magnetic field.

** Not credible.

Well, maybe it is. The technology of the day was weak on switchmode
circuitry, and I\'ve seen exploded-wire failures (in metal cases, just file
off the lid and aim a microscope at the problem).

The dI/dt limit on a few millimeters of wire was due to magnetic
forces, skin effect, and shockwve (ultrasonic) stresses. Not the
terrestrial field, the current-induced field. Unlike thermal failure,
there\'s no blob of melt at the end of the exploded wire... it\'s just
GONE.

I recently tested some pcb mounted fuses to see how well they would
survive the startup surge of a dc input power supply where a large
capacitor was charged up very fast.
After around 100,000 startups the fuse would fail from metal fatigue
at the points where the fuse wire is welded to the connection wires.
The peak current was around 30 to 50 times the rated current of the fuse
for a few tens of microseconds.
I also repeated the test with a powerful magnet on top of the fuse
to see whether this made any difference. It did not.

I think the fuse wire \"instantaneously\" lengthened, thereby changing
the angle of the wire at the terminations, effectively bending it
backwards and forwards slightly each time. The geometry is probably
very similar to that of bond wires in a power transistor.

John

Not all thermal effects are slow, mainly just thermal conduction. I
have sampling scope data showing a metal bolometer attached to one of my
SOI waveguide-coupled nanoantennas, with 20 ps pulses from a big
lamp-pumped laser. The measured resistance changed in like 30 ps.

There are other slower processes, such as the thermalization of
different energy reservoirs in loosely coupled systems--ion vs. electron
temperatures in low-pressure plasmas, for instance--but in solids heat
dissipation and thermal expansion is pretty fast. (Of course the actual
motion of the heated metal happens on acoustic time scales, i.e. much
slower.)

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

 

[Phil Hobbs]

On 2020-07-29 10:44, jlarkin@highlandsniptechnology.com wrote:

On Wed, 29 Jul 2020 03:38:09 -0700 (PDT), jrwalliker@gmail.com wrote:

On Tuesday, 28 July 2020 17:49:52 UTC+1, whit3rd wrote:
On Tuesday, July 28, 2020 at 4:17:51 AM UTC-7, Phil Allison wrote:
Martin Brown wrote:
--------------------


I recall an infamous PWM X10 power amplifier designed in 1964 by
idiosyncratic but brilliant Ivor Catt

The thin bond out wires eventually failed due to vibration from the high
currents interacting with the Earth\'s magnetic field.

** Not credible.

Well, maybe it is. The technology of the day was weak on switchmode
circuitry, and I\'ve seen exploded-wire failures (in metal cases, just file
off the lid and aim a microscope at the problem).

The dI/dt limit on a few millimeters of wire was due to magnetic
forces, skin effect, and shockwve (ultrasonic) stresses. Not the
terrestrial field, the current-induced field. Unlike thermal failure,
there\'s no blob of melt at the end of the exploded wire... it\'s just
GONE.

I recently tested some pcb mounted fuses to see how well they would
survive the startup surge of a dc input power supply where a large
capacitor was charged up very fast.
After around 100,000 startups the fuse would fail from metal fatigue
at the points where the fuse wire is welded to the connection wires.
The peak current was around 30 to 50 times the rated current of the fuse
for a few tens of microseconds.
I also repeated the test with a powerful magnet on top of the fuse
to see whether this made any difference. It did not.

I think the fuse wire \"instantaneously\" lengthened, thereby changing
the angle of the wire at the terminations, effectively bending it
backwards and forwards slightly each time. The geometry is probably
very similar to that of bond wires in a power transistor.

John

We don\'t use surface-mount fuses. We tested some that would fail at
currents within their hold ratings. They depend too much on PCB
thermals and maybe on luck.

Ditto surface-mount polyfuses. We use leaded ones successfully.

And PolyZens have all gone away. *sob*

\"Silicon fuses\", \"E-fuses\", have their own problems.

In general, protecting gear from wall-warts is tricky.

Yup.

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

 

[Phil Hobbs]

On 2020-07-28 16:41, Rich S wrote:

On Tuesday, July 28, 2020 at 3:04:24 PM UTC, Anthony Stewart wrote:
On Tue. 28 Jul.-20 7:59 a.m., jurb6006@gmail.com wrote:
We use one fet that\'s claimed to be able to dissipate 1000 watts

The highest dissipation I have ever seen for a transistor was 625 watts. Made by Westinghouse and slow as hell. I am sure it it no longer available. To even find the part number would take me days in my ancient spec book.

If you can directly cool the die you can get quite a bit. The limiting factor other than current ad voltage is the thermal resistance to the case. So maybe you could take a 2N3055 and drill two holes in it and pump coolant right inside of it.

Of course there are no guarantees.


I have used snow,ice, a toaster, a power diode and 120Vac to start my
car in the dead of winter at -30\'C in the 80\'s

I would connect Neutral to Vbat Gnd and Line thru the toaster to the
diode in a jar of water with ice in it to start the car in few minutes
when the battery was in poor condition. The peak current was enough to
charge the battery.

Did you prepare breakfast (toast) at the same time?

When I was a kid I used a teakettle as the ballast in an arc lamp made
from the carbons of two D cells arranged inside a large Colman\'s mustard
tin. That sucker was _bright_. My parents had lots of stomach lining.

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

 

[Lasse Langwadt Christensen]

torsdag den 30. juli 2020 kl. 00.25.55 UTC+2 skrev Phil Hobbs:

On 2020-07-28 16:41, Rich S wrote:
On Tuesday, July 28, 2020 at 3:04:24 PM UTC, Anthony Stewart wrote:
On Tue. 28 Jul.-20 7:59 a.m., jurb6006@gmail.com wrote:
We use one fet that\'s claimed to be able to dissipate 1000 watts

The highest dissipation I have ever seen for a transistor was 625 watts. Made by Westinghouse and slow as hell. I am sure it it no longer available. To even find the part number would take me days in my ancient spec book.

If you can directly cool the die you can get quite a bit. The limiting factor other than current ad voltage is the thermal resistance to the case. So maybe you could take a 2N3055 and drill two holes in it and pump coolant right inside of it.

Of course there are no guarantees.


I have used snow,ice, a toaster, a power diode and 120Vac to start my
car in the dead of winter at -30\'C in the 80\'s

I would connect Neutral to Vbat Gnd and Line thru the toaster to the
diode in a jar of water with ice in it to start the car in few minutes
when the battery was in poor condition. The peak current was enough to
charge the battery.

Did you prepare breakfast (toast) at the same time?


When I was a kid I used a teakettle as the ballast in an arc lamp made
from the carbons of two D cells arranged inside a large Colman\'s mustard
tin. That sucker was _bright_. My parents had lots of stomach lining.

https://youtu.be/HTJKksnurS0

 

[Jeff Layman]

On 28/07/2020 09:29, Martin Brown wrote:

On 26/07/2020 17:51, Phil Hobbs wrote:

I recall an infamous PWM X10 power amplifier designed in 1964 by
idiosyncratic but brilliant Ivor Catt for Clive Sinclair that was class
D using seriously under rated germanium transistors in hard saturation.
The thin bond out wires eventually failed due to vibration from the high
currents interacting with the Earth\'s magnetic field. The power rating
of the transistors was not actually exceeded but their tolerance of
brutal currents flowing in them was. It wasn\'t a commercial success but
it put him on the map. I thought I had imagined it until I found this OU
document where the X10 gains honourable mention on p13 although being a
marketing course it doesn\'t mention the technical cause of failure.

It actually worked pretty well for the price but died soon enough if
used at anything more than 25% of rated power output that they had a lot
of very unhappy customers. The later X20 did OK and they went on to
develop quite a hobby following with various kits and very early cheap
calculators and digital watches the size of a matchbox!

The Sinclair Z12 was no better - in fact probably worse. I bought one in
late 1968 and soon found out it went into a sort of motorboating
distortion after barely 30 minutes. Well, what would you expect with
reject germanium power transistors (were they OC26?) and a totally
inadequate heat sink?

http://www.sinclairql.net/downloads/1986_Sinclair_QL_microcomputer_case_study_by_Godfrey_Boyle_OpenUniversity_T362_Block_2_Marketing-SCN04-SQPP.pdf

Top left of page 13 describes the thing. Later on Catt was better known
for advocating wafer scale integration and railing against relativity.

The QL was doomed by Sir Clive\'s insistence that it used his own tape
microdrives (so prone to jamming) rather than floppy disks. The rather
tacky keyboard didn\'t help but was much better than a ZX80 one! Apart
from that it was a good machine with a 68008 CPU and a Basic dialect
that was well ahead of its time. Sadly it never became very popular.

I should have known better after my experience with the Z12, but I
bought a QL in September 1984. It actually worked quite well, especially
with the extra 500k of ram I bought. But, of course, the famous Sinclair
reliability issue came into play when one of the Microdrives failed -
the one that loaded the OS, of course, so the machine became unusable.
Not long after I started using an IBM desktop at work (386 running DOS
3.3), and soon realised that was the future. I still have the QL in the
loft.

--

Jeff

 

[Tom Gardner]

On 30/07/20 15:32, Jeff Layman wrote:

On 28/07/2020 09:29, Martin Brown wrote:
On 26/07/2020 17:51, Phil Hobbs wrote:

I recall an infamous PWM X10 power amplifier designed in 1964 by
idiosyncratic but brilliant Ivor Catt for Clive Sinclair that was class
D using seriously under rated germanium transistors in hard saturation.
The thin bond out wires eventually failed due to vibration from the high
currents interacting with the Earth\'s magnetic field. The power rating
of the transistors was not actually exceeded but their tolerance of
brutal currents flowing in them was. It wasn\'t a commercial success but
it put him on the map. I thought I had imagined it until I found this OU
document where the X10 gains honourable mention on p13 although being a
marketing course it doesn\'t mention the technical cause of failure.

It actually worked pretty well for the price but died soon enough if
used at anything more than 25% of rated power output that they had a lot
of very unhappy customers. The later X20 did OK and they went on to
develop quite a hobby following with various kits and very early cheap
calculators and digital watches the size of a matchbox!

The Sinclair Z12 was no better - in fact probably worse. I bought one in late
1968 and soon found out it went into a sort of motorboating distortion after
barely 30 minutes. Well, what would you expect with reject germanium power
transistors (were they OC26?) and a totally inadequate heat sink?

Sinclair\'s reputation was that his audio products were dreadful;
there were many many bad stories in the letter columns of the
electronics magazines.

Even though I would have been perfectly capable of assembling
it myself, that reputation made me buy an assembled and
working Sinclair Scientific calculator.

It still works, and is as accurate as a slide rule - which
is a remarkable achievement considering /everything/ was
contained in a 320 instruction ROM.
Ken Shirriff has an online simulator, which is worth a look.

 

[server]

On Thursday, 30 July 2020 15:51:51 UTC+1, Tom Gardner wrote:

Sinclair\'s reputation was that his audio products were dreadful;
there were many many bad stories in the letter columns of the
electronics magazines.

The IC10 claimed to be a 10W amplifier, but it was a rebadged Plessey
chip rated by Plessey at something like 3.5W.

Even though I would have been perfectly capable of assembling
it myself, that reputation made me buy an assembled and
working Sinclair Scientific calculator.

It still works, and is as accurate as a slide rule - which
is a remarkable achievement considering /everything/ was
contained in a 320 instruction ROM.
Ken Shirriff has an online simulator, which is worth a look.

I had one of those. It was horrible as it could sometimes give
results that were much less accurate than a slide rule.
I was caught out when dividing a number by the difference between
two similar numbers. I think the error was several percent. This
was a problem because I didn\'t notice at the time.

John

 

[Tom Gardner]

On 30/07/20 23:27, jrwalliker@gmail.com wrote:

On Thursday, 30 July 2020 15:51:51 UTC+1, Tom Gardner wrote:

Sinclair\'s reputation was that his audio products were dreadful;
there were many many bad stories in the letter columns of the
electronics magazines.

The IC10 claimed to be a 10W amplifier, but it was a rebadged Plessey
chip rated by Plessey at something like 3.5W.

Even though I would have been perfectly capable of assembling
it myself, that reputation made me buy an assembled and
working Sinclair Scientific calculator.

It still works, and is as accurate as a slide rule - which
is a remarkable achievement considering /everything/ was
contained in a 320 instruction ROM.
Ken Shirriff has an online simulator, which is worth a look.

I had one of those. It was horrible as it could sometimes give
results that were much less accurate than a slide rule.
I was caught out when dividing a number by the difference between
two similar numbers. I think the error was several percent. This
was a problem because I didn\'t notice at the time.

I seem to remember something similar, but when I tried
what I remembered, the answer was correct.

If you don\'t have a working Scientific to hand, you could
try to repeat the experience on this notable emulator
http://files.righto.com/calculator/sinclair_scientific_simulator.html

 

[server]

On Friday, 31 July 2020 08:44:06 UTC+1, Tom Gardner wrote:

On 30/07/20 23:27, jrwalliker@gmail.com wrote:
On Thursday, 30 July 2020 15:51:51 UTC+1, Tom Gardner wrote:

Sinclair\'s reputation was that his audio products were dreadful;
there were many many bad stories in the letter columns of the
electronics magazines.

The IC10 claimed to be a 10W amplifier, but it was a rebadged Plessey
chip rated by Plessey at something like 3.5W.

Even though I would have been perfectly capable of assembling
it myself, that reputation made me buy an assembled and
working Sinclair Scientific calculator.

It still works, and is as accurate as a slide rule - which
is a remarkable achievement considering /everything/ was
contained in a 320 instruction ROM.
Ken Shirriff has an online simulator, which is worth a look.

I had one of those. It was horrible as it could sometimes give
results that were much less accurate than a slide rule.
I was caught out when dividing a number by the difference between
two similar numbers. I think the error was several percent. This
was a problem because I didn\'t notice at the time.

I seem to remember something similar, but when I tried
what I remembered, the answer was correct.

If you don\'t have a working Scientific to hand, you could
try to repeat the experience on this notable emulator
http://files.righto.com/calculator/sinclair_scientific_simulator.html

I would first have to find my physics practical lab notes from 1977.
They were almost certainly binned a very long time ago.


John

 

[Tom Gardner]

On 31/07/20 14:33, jrwalliker@gmail.com wrote:

On Friday, 31 July 2020 08:44:06 UTC+1, Tom Gardner wrote:
On 30/07/20 23:27, jrwalliker@gmail.com wrote:
On Thursday, 30 July 2020 15:51:51 UTC+1, Tom Gardner wrote:

Sinclair\'s reputation was that his audio products were dreadful;
there were many many bad stories in the letter columns of the
electronics magazines.

The IC10 claimed to be a 10W amplifier, but it was a rebadged Plessey
chip rated by Plessey at something like 3.5W.

Even though I would have been perfectly capable of assembling
it myself, that reputation made me buy an assembled and
working Sinclair Scientific calculator.

It still works, and is as accurate as a slide rule - which
is a remarkable achievement considering /everything/ was
contained in a 320 instruction ROM.
Ken Shirriff has an online simulator, which is worth a look.

I had one of those. It was horrible as it could sometimes give
results that were much less accurate than a slide rule.
I was caught out when dividing a number by the difference between
two similar numbers. I think the error was several percent. This
was a problem because I didn\'t notice at the time.

I seem to remember something similar, but when I tried
what I remembered, the answer was correct.

If you don\'t have a working Scientific to hand, you could
try to repeat the experience on this notable emulator
http://files.righto.com/calculator/sinclair_scientific_simulator.html

I would first have to find my physics practical lab notes from 1977.
They were almost certainly binned a very long time ago.

When, around that time, I had to do lab calculations, I
tended to go to the university library and use the HP35.
It was carefully bolted to the librarian\'s desk

Now I have one of my own, in a hand-crafted hardwood box.