I've mentioned in the past that I had done a design for a very low
cost DC-DC converter but that I was reluctant to share details until
I had published it. I'm happy to say that I finally got off my
rear-end and did just that.
8-32 Vdc input, 5Vdc @ 50 mA output. Efficiency not measured but
probably around 80% when loaded. Much worse when un-loaded because
unused current is dissipated in a zener diode. This was absolutely
NOT a concern for me because the load was essentially constant.
Very low cost. I'm using this in a couple of products and our
component cost is about $0.60 including the common-mode choke that
I'm using as a transformer. We purchased several thousand pieces of
that choke for about US $0.25 each several years ago.
Really strange oscillator circuit based on one section of a dual or
quad comparitor. I've used both LM393 and LM339 comparitors and they
work really well. Oscillator relies on the open-collector output of
the comparitor.
One gripe I have with many circuits published is that while a method for
characterizing a choke for possible used was provided (that's a lot more
than most authors do), it irritates me when a mystery part, such as "FL601"
is used. How many chokes have to be bought before one that is useable can be
identified? A mystery part is a mystery part. I think EDN, etc should
require authors to provide manufacturer, part number, and values on all
circuits/parts that it publishes. It is seem like good policy.
Dwayne Reid wrote:
> Its in Electronic Design's Ideas For Design column as linked here:
>
<electronicdesign.com/article/ideas-for-design/build_your_own_ultra_l
ow_cost_isolated_dc_dc_converter.aspx>http://electronicdesign.com/article/id
eas-for-design/build_your_own_ultra_low_cost_isolated_dc_dc_converter.aspx
"Bad Request (Invalid URL)"
How about a more reasonable link to just the schematic PDF?
> I've mentioned in the past that I had done a design for a very low
> cost DC-DC converter but that I was reluctant to share details until
> I had published it. I'm happy to say that I finally got off my
> rear-end and did just that.
>
> Its in Electronic Design's Ideas For Design column as linked here:
> <<http://electronicdesign.com/article/ideas-for-design/build_your_own_ultra_low_cost_isolated_dc_dc_converter.aspx>http://electronicdesign.com/article/ideas-for-design/build_your_own_ultra_low_cost_isolated_dc_dc_converter.aspx>
>
> A few details:
>
> 8-32 Vdc input, 5Vdc @ 50 mA output. Efficiency not measured but
> probably around 80% when loaded. Much worse when un-loaded because
> unused current is dissipated in a zener diode. This was absolutely
> NOT a concern for me because the load was essentially constant.
>
> Very low cost. I'm using this in a couple of products and our
> component cost is about $0.60 including the common-mode choke that
> I'm using as a transformer. We purchased several thousand pieces of
> that choke for about US $0.25 each several years ago.
>
> Really strange oscillator circuit based on one section of a dual or
> quad comparitor. I've used both LM393 and LM339 comparitors and they
> work really well. Oscillator relies on the open-collector output of
> the comparitor.
>
> Hope this is able to help someone.
Looks very interesting. Why was 2N5401 selected for Q1 (as opposed to a FET,
for example)? How noisy is the circuit?
Olin Lathrop wrote:
>> Its in Electronic Design's Ideas For Design column as linked here:
>>
> <electronicdesign.com/article/ideas-for-design/build_your_own_ultra_l
> ow_cost_isolated_dc_dc_converter.aspx>http://electronicdesign.com/article/id
> eas-for-design/build_your_own_ultra_low_cost_isolated_dc_dc_converter.aspx
>
> "Bad Request (Invalid URL)"
>
> How about a more reasonable link to just the schematic PDF?
Em 9/4/2010 17:05, Olin Lathrop escreveu:
> Dwayne Reid wrote:
>
>> Its in Electronic Design's Ideas For Design column as linked here:
>>
>>
> <electronicdesign.com/article/ideas-for-design/build_your_own_ultra_l
> ow_cost_isolated_dc_dc_converter.aspx>http://electronicdesign.com/article/id
> eas-for-design/build_your_own_ultra_low_cost_isolated_dc_dc_converter.aspx
>
> "Bad Request (Invalid URL)"
>
> How about a more reasonable link to just the schematic PDF?
>
Olin, it should be a problem with your e-mail client. Every link worked
OK in the message I received.
Isaac
__________________________________________________
Fale com seus amigos de graça com o novo Yahoo! Messenger http://br.messenger.yahoo.com/
> Dwayne Reid wrote:
>> Its in Electronic Design's Ideas For Design column as linked here:
>>
> <electronicdesign.com/article/ideas-for-design/build_your_own_ultra_l
> ow_cost_isolated_dc_dc_converter.aspx>http://electronicdesign.com/article/id
> eas-for-design/build_your_own_ultra_low_cost_isolated_dc_dc_converter.aspx
>
> "Bad Request (Invalid URL)"
>
> How about a more reasonable link to just the schematic PDF?
>
>
> ********************************************************************
> Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
> (978) 742-9014. Gold level PIC consultants since 2000.
> Dwayne Reid wrote:
>
>> Its in Electronic Design's Ideas For Design column as linked here:
>>
>>
> <electronicdesign.com/article/ideas-for-design/build_your_own_ultra_l
> ow_cost_isolated_dc_dc_converter.aspx>http://electronicdesign.com/article/id
> eas-for-design/build_your_own_ultra_low_cost_isolated_dc_dc_converter.aspx
>
> "Bad Request (Invalid URL)"
>
> How about a more reasonable link to just the schematic PDF?
>
Olin, it should be a problem with your e-mail client. Every link worked
OK in the message I received.
Isaac
If you look at the link above you will see that it is doubled. Actually it
is in the form:
<link>link
with two copies of the link data.
I had to manually edit it to one occurrence before it would work for me.
> I've mentioned in the past that I had done a design
> for a very low cost DC-DC converter but that I was
> reluctant to share details until I had published it.
> I'm happy to say that I finally got off my rear-end
> and did just that.
>
> Its in Electronic Design's Ideas For Design column
> as linked here:
Why not just make some PIC to drive your choke (through a FET)? Yes, a
FET, a resistor + zener for the PIC, but you don't need a bunch of
other parts. I believe I've seen something by Olin related to that.
"Marechiare" wrote:
>> I've mentioned in the past that I had done a design
>> for a very low cost DC-DC converter but that I was
>> reluctant to share details until I had published it.
>> I'm happy to say that I finally got off my rear-end
>> and did just that.
>>
>> Its in Electronic Design's Ideas For Design column
>> as linked here:
>
> Why not just make some PIC to drive your choke (through a FET)? Yes, a
> FET, a resistor + zener for the PIC, but you don't need a bunch of
> other parts. I believe I've seen something by Olin related to that.
IIRC the comparator can be had for 1/3 the cost of the 10F. The rest are
penny parts.
Vitaliy wrote:
> "Marechiare" wrote:
>
>>> I've mentioned in the past that I had done a design
>>>
>>>
>> Why not just make some PIC to drive your choke (through a FET)? Yes, a
>>
>>
> IIRC the comparator can be had for 1/3 the cost of the 10F. The rest are
> penny parts.
>
> Vitaliy
>
>
or even do it without the comparator.
google single transistor buck inverter
and single transistor forward converter
On Sat, 10 Apr 2010 00:50:51 +0100, "Michael Watterson" said:
> or even do it without the comparator.
>
> google single transistor buck inverter
> and single transistor forward converter
>
> If it's low power & zener regulated you don't need any control circuit.
> http://www.laserfaq.org/sam/ssinv1.pdf
> or
> http://www.laserfaq.org/sam/ssinv1.pdf
> (These are for flash guns, but with a 1:1 transformer and zener. they
> can work as per original circuit.)
>
> I see no value to the comparator compared with simple inverter if you
> use a zener on output rather than feedback
IMO, there are some subtleties to Dwayne's circuit that make it pretty
nice. Current mode feedback is good. A 3:1 input voltage range is hard
to do cleanly with just a transistor or two - especially if you expect
to make more than one of them without playing around with resistor
values :)
>>> I've mentioned in the past that I had done a design
>>> for a very low cost DC-DC converter but that I was
>>> reluctant to share details until I had published it.
>>> I'm happy to say that I finally got off my rear-end
>>> and did just that.
>>>
>>> Its in Electronic Design's Ideas For Design column
>>> as linked here:
>>
>> Why not just make some PIC to drive your choke
>> (through a FET)? Yes, a FET, a resistor + zener for
>> the PIC, but you don't need a bunch of other parts.
>>I believe I've seen something by Olin related to that.
>
> IIRC the comparator can be had for 1/3 the cost of
> the 10F. The rest are penny parts.
In small quantities that does not matter, in large quantities the
difference should be much less. I think, In large quantities PIC10F
should be sort of 10 cents. The comparator plus that pile of resistors
and capacitors would total more than single PIC, I'd suggest. Besides,
extra PCB space, parts parameters control, extra assembling costs,
lack of functional flexibility, reliability issues with the dirt cheap
components, etc. I'd rather adapt Olin's design.
>Looks very interesting. Why was 2N5401 selected for Q1 (as opposed to a FET,
>for example)? How noisy is the circuit?
Several reasons for using a bipolar transistor instead of a
MOSFET. Cost was one major reason. It was hard to find a decent
P-channel MOSFET when this design was done. Filedates say that this
was designed in 2004.
Another major concern was ensuring that the FET would turn off
quickly. With only an open-collector output stage, I'd have had to
add another transistor so as to discharge the gate capacitance.
The final reason for not using a MOSFET was the supply voltage. The
circuit was specified to operate up to 32 Vdc. I would have had to
add some sort of clamp to ensure that the MOSFET gate voltage did not
exceed 18V or so.
The combination of an inexpensive PNP bipolar transistor with a 100R
E-B resistor satisfied all of the above concerns.
As far as noise (EMC) goes, I truly don't know. We weren't required
to do emissions testing, so we didn't. It certainly can't be any
worse than a typical National Semiconductor Simple Switcher, though.
Keep in mind that this was designed as a low energy supply. Its
first use was in the input stage of DMX lighting controllers, where
it was used to supply power to the RS-422 receiver chip and the
opto-isolator used to get the data into the actual controller. It
has worked admirably in that application since 2004.
At 01:53 PM 4/9/2010, Richard Pytelewski wrote:
>Dwayne:
>
>One gripe I have with many circuits published is that while a method for
>characterizing a choke for possible used was provided (that's a lot more
>than most authors do), it irritates me when a mystery part, such as "FL601"
>is used. How many chokes have to be bought before one that is useable can be
>identified? A mystery part is a mystery part.
Keep in mind that I simply found a low cost supplier of common-mode
chokes and told him: "Send me samples of whatever you have the most
of right now". I then worked from those samples.
That part number "FL601" was the ONLY documentation I had on the part.
But: it didn't much matter. Every common-mode choke I tried was able
to couple energy from one winding to another. All I had to do was
find out what frequency worked best and what the maximum duty cycle
had to limited to for each particular choke. Then I used LT Spice to
get the oscillator to do what I wanted, then confirmed actual
operation on the bench.
By the way, I have a zip file of several versions of the LT Spice
files for the oscillator section if anyone wants to play with
simulation before actually trying it out. 14 different variations,
ranging from 500Hz up to 400 KHz. Please ask me privately if you
want a copy. I will NOT respond to requests made on the list.
I don't know how well the files will work in simulation packages
other than LT Spice (also known as SwitcherCAD) but LT Spice is free
and VERY well maintained. It should be no hardship to download and
install it if you don't already have it.
>In small quantities that does not matter, in large quantities the
>difference should be much less. I think, In large quantities PIC10F
>should be sort of 10 cents. The comparator plus that pile of resistors
>and capacitors would total more than single PIC, I'd suggest. Besides,
>extra PCB space, parts parameters control, extra assembling costs,
>lack of functional flexibility, reliability issues with the dirt cheap
>components, etc. I'd rather adapt Olin's design.
No problem! If that works better for you, that's what you should do.
> >> Why not just make some PIC to drive your choke
> >> (through a FET)? Yes, a FET, a resistor + zener for
> >> the PIC, but you don't need a bunch of other parts.
> >>I believe I've seen something by Olin related to that.
> >
> > IIRC the comparator can be had for 1/3 the cost of
> > the 10F. The rest are penny parts.
>
> In small quantities that does not matter, in large quantities the
> difference should be much less. I think, In large quantities PIC10F
> should be sort of 10 cents.
If ANYONE has any indication that a 10F can be had for 10 cents in any
volume under1 million I'd be please to know of it right now.
It's been a while since I looked but last time the cheapest 10F seemed
to come down to about 30c and then it was "please ask". I had imagined
that 20-25 c would be about the best doable.
No time to comment in vast detail (consider yourself blessed) just now
BUT it seems that the voltage on the output snubber is a direct analog
of the output voltage so could be used for regulation. eg just turn on
a high side bipolar and use that to back off oscillator.
PIC etc is a valid way to do this BUT comparator does not need
programming, has good sink drive, switches faster than the cheapest
smps ICs, is multisource available, ... .
I'd consider a TL431 or TLV431 as a reference. Probably don't need 1st
buffer comparator then - use for voltage regulation or other use.
TL431 is about $US0.015 in modest volume in Asia !!! :-). TLV431 has
much lower minimum regulation current leading to potential for low
shutdown power.
There is still a place for custom smps based on cheap std ics. I did
vaguely similar some years ago with a 74C14 hex inverter driver to
make a supply that would regulate at no load with all up current about
100 uA including 2 x LDOs.
But note that there are quite a number of Asian sourced ICs that would
probably do this task and perhaps as cheap. eg CE9908 as a starting
point. These are made for the cellphone charger market and similar
boost applications.
Isolated output is a significant feature.
The comparator plus that pile of resistors
> and capacitors would total more than single PIC, I'd suggest. Besides,
> extra PCB space, parts parameters control, extra assembling costs,
> lack of functional flexibility, reliability issues with the dirt cheap
> components, etc. I'd rather adapt Olin's design.
> -
On Apr 10, 2010, at 4:54 AM, Russell McMahon wrote:
> There is still a place for custom smps based on cheap std ics.
> Isolated output is a significant feature.
I for one would very much like to see a PoE (Power over Ethernet)
circuit(s) capable of supplying 0.5 to 2W isolated power. (though I
suspect that the spec is just too complex; seems the whole "internet
of things" meme came later and PoE is mostly so you can provide 10+ W
to expensive IP phones. Sigh.) This circuit looks like it would
come close for the regulation part. What limits the input voltage to
30V? Just the max voltage of the 393 ?
> I'd consider a TL431 or TLV431 as a reference. Probably don't need 1st
> buffer comparator then - use for voltage regulation or other use.
> TL431 is about $US0.015 in modest volume in Asia !!! :-). TLV431 has
> much lower minimum regulation current leading to potential for low
> shutdown power.
There are a lot of regulated isolated DC/DC converter using TL431 and
an opto for the feedback, many of them being a flyback converter.
In my previous job, I've seen many 555 timer controlled flyback
converter with TL431/Opto as the feedback.
> There is still a place for custom smps based on cheap std ics. I did
> vaguely similar some years ago with a 74C14 hex inverter driver to
> make a supply that would regulate at no load with all up current about
> 100 uA including 2 x LDOs.
>
I've done similar things for a NAMUR sensor. The isolated (unregulated)
push-pull converter is driven by a CMOS multi-vibrator. The whole
circuitry (including a PIC, two LED, one of them flashing to indicate
it is alive, and level sensing circuitry based on a bridge) needs to
consume lower than 1.2mA if not on. When ON, the other LED
lights up and the whole circuitry need to consume between 2.1mA
to 4mA. http://www.microchip.com/forums/fb.aspx?m=427267
And for the same family, there is a DC version where I use an even
cheaper push-pull based on two transistors.
Then there is this universal AC/DC version (20...250 V AC/DC), there
is no PWM IC to be used for this range. So again I have to built
the DC/DC converter based on Transistors to regulate
the voltage down to 9V for relay driving. Then an isolated push-pull
will do the job of providing power to the field side.
> >What limits the input voltage to 30V? Just the max voltage of the 393 ?
> Yep. Actually, the '393 is rated to 35V, I spec 32V as max just for
> my usual 10% safety margin.
> I suspect that some '393s will go past 40V but not in any product
> that I would ship to a customer.
> The max voltage limit is both the max Vcc on the chip as well as the
> max allowable voltage on the output pin of the 2nd comparitor.
You can extend the voltage indefinitely by running the 339 from a
lower voltage supply and by adding a driver or voltage limiting
circuitry between the opamp and Q1. The easy way is to add and NPN
driven by U2 pin 7 with a resistor from its collector to Q2 base.
You'll need to swap U2 pins 5 & 6 due to the inversion from the
transistor. A small amount of playing may be needed with component
values. OR adding a zener diode or other offset circuitry would allow
higher voltages to be used.
For much higher voltages U2 could be started by a bleed resistor from
HV. to supply it's power supply and then driven by a 3rd winding -
unlikely to be popular given the effort gone to to get a COTS 2
winding inductor initially.
If an eg TL431 is used to drive U2 pin 6 (in original cct), then the
flyback voltage on C1 can be used with the now unused comparator to
provide voltage regulation. This wont be as tight as if the secondary
was used (due to leakage inductance, spikes and other real world joys)
but much better than no voltage regulation at all.
Pushing your luck you can try this circuit with uncoupled inductors
and a largish capacitor between the input and output sides (location
left as an excercise etc ...) thus producing a SEPIC converter which
really needs to be controlled.
'William Chops" Westfield ' <KILLspamwestfwKILLspammac.com wrote:
> I for one would very much like to see a PoE (Power over Ethernet)
> circuit(s) capable of supplying 0.5 to 2W isolated power.
That is easily available. I fail to see the problem. For example,
Coilcraft makes POE transformers for three different power ranges, the
lowest being 3W if I remember right.
No matter what you do with the switcher though, you still have to do the POE
negotiation. There are chips from various vendors, including TI and
National for that. One National chip I saw used recently included the
swither controller and the POE negotiation in the same chip.
Em 11/4/2010 11:21, Olin Lathrop escreveu:
> 'William Chops" Westfield ' <RemoveMEwestfwTakeThisOuTmac.com wrote:
>
>> I for one would very much like to see a PoE (Power over Ethernet)
>> circuit(s) capable of supplying 0.5 to 2W isolated power.
>>
> That is easily available. I fail to see the problem. For example,
> Coilcraft makes POE transformers for three different power ranges, the
> lowest being 3W if I remember right.
>
> No matter what you do with the switcher though, you still have to do the POE
> negotiation. There are chips from various vendors, including TI and
> National for that. One National chip I saw used recently included the
> swither controller and the POE negotiation in the same chip.
>
I use the LM5070 from National. It needs some external passives, a
transformer, a power MOSFET, a TL431 and an opto-coupler.
Isaac
__________________________________________________
Fale com seus amigos de graça com o novo Yahoo! Messenger http://br.messenger.yahoo.com/
>> I for one would very much like to see a PoE (Power over Ethernet)
>> circuit(s) capable of supplying 0.5 to 2W isolated power.
>
> That is easily available. I fail to see the problem. For example,
> Coilcraft makes POE transformers for three different power ranges, the
> lowest being 3W if I remember right.
I left out the phrase "low cost", since I was trying to stay on-topic
with the original subject line...
> I use the LM5070 from National. It needs some external passives, a
> transformer, a power MOSFET, a TL431 and an opto-coupler.
By the time you pay $3 for the specialty controller chip (Q100 from
digikey), and $4 for the specialty transformer (Q100 from mouser; DK's
MOQ is 200; neither actually stocks the lower power transformers), the
whole thing isn't looking very attractive...
Can you use the LM5070 without an external MOSFET? It sounds like the
internal switches (800mA gate driver) should be sufficient for low
power supplies, but they don't document using it that way, and I don't
see how you'd set up the whole current sense feedback setup with such
a setup... The example circuit in the lm5070 datasheet is
disturbingly more complex than the simplified block diagram :-(
'William Chops" Westfield ' <spamBeGonewestfwspamBeGonemac.com wrote:
> By the time you pay $3 for the specialty controller chip (Q100 from
> digikey), and $4 for the specialty transformer (Q100 from mouser; DK's
> MOQ is 200; neither actually stocks the lower power transformers), the
> whole thing isn't looking very attractive...
You are making this too complicated and needlessly expensive. One of my
side projects right now is experimenting to make a reasonably affordable
isolated supply that takes a fairly wide input power range. I want
something I can plunk down for isolated power from a 48V DC bus, but that
can also work just as well with as low as 12V DC in minimum, preferably
less.
This is work in progress. I put the current snapshot at http://www.embedinc.com/temp/isol1.pdf. The transformer is from CoilCraft,
and is meant for flyback operation for POE applications. It costs well
under $1 in quantities. The PIC 16F616 is only because that's what I had in
stock of that family of PICs that contains the power supply PWM module with
shutdown. Eventually I expect to use a 12F615, or maybe a 12HV615. Samples
are on order, but for now I can test just fine with the 16F616 because it's
the same thing with 14 pins instead of 8. The 12F615-I/SN costs $.57 each
for 1000.
As I said, this is work in progress and a testbed. Even though the point is
to create a isolated supply, I have both grounds connected together to make
probing around easier. It should be obvious that the two grounds could be
separated. This is built on a ReadyBoard-01, so I'm using the convenience
of having known regulated 5V available. Vout is intended to be one diode
drop above the regulated output. It would normally be the input to a LDO,
with the right side of R4 connected to the output. In this case I used the
existing 5V line instead of putting the LDO on the test board.
D1 does nothing in the current circuit. It is left over from earlier
experiments. I left it on the schematic because it is still physically on
the test board.
The firmware strategy is to use the hardware PWM with shutdown to generate
the transformer pulses. The PWM duty cycle will be optimum for the
particular input voltage. The NSHUT line is driven low by the opto when the
output is above its regulation threshold. This shuts down the PWM output
automatically without firmware intervention. The foreground code
periodically polls the A/D and adjusts the PWM duty cycle according to the
input voltage. This all works fine so far, although the best relationship
between input voltage and duty cycle still has to be determined. It's easy
to change the duty cycle on the fly, but not the PWM period. Right now the
PWM is set to a fixed 100KHz, so at low voltages the raw pulse width based
on transformer saturation alone is too long without the duty cycle fraction
being considered. That consideration isn't in the code yet.
Eventually this needs to create a 5V supply on the input side too. I plan
on using interrupts and the comparator input and internal absolute voltage
reference for that. The bias winding of the transformer is tempting, but
I'm not sure it can be used without assuming a minimum load on the output,
which I don't want to do in this application. I may use a separate inductor
with a simple PNP high side switch.
So in brief, the hardware PWM will generate and regulate the isolated
supply, interrupts with comparator and internal voltage reference the local
supply, and the foreground code will read the input voltage and adjust the
PWM duty cycle. It will probably also leave information around for the
local supply pulse width.
>'William Chops" Westfield ' <TakeThisOuTwestfwEraseMEspam_OUTmac.com wrote:
>> By the time you pay $3 for the specialty controller chip (Q100 from
>> digikey), and $4 for the specialty transformer (Q100 from mouser; DK's
>> MOQ is 200; neither actually stocks the lower power transformers), the
>> whole thing isn't looking very attractive...
On the subject of DC/DC transformers - I recently did some hunting for a cheap isolated RS485
solution - I ended up with AD's ADM2487 and this Murata transformer :
Introduction
This application note describes the STEVAL-TSP001V1 demonstration board for the
evaluation of a power over Ethernet (PoE) system used to transmit
electrical power, along
with data, to remote devices over a standard twisted-pair cable in an
Ethernet network.
A complete power solution for Ethernet-connected powered devices is
presented. The
power supply fully complies with IEEE 802.3-2005 PoE specifications
and delivers the rated
output from any compliant source.
Example outputs of 3.3 V and 12 V are given in this document, but
other requirements can
easily be met by implementing small changes.
Key features
â– Low profile, small size: 5"w x 1.5"d x 0.65"h (0.5" except RJ-45
Ethernet connectors)
â– Complies with all IEEE 802.3-2005 (IEEE 802.3af) Power over Ethernet
specifications
â– Respects source limitations
â– Does not pollute power source
â– 1500 V isolated outputs eliminate ground loops
â– Highest possible economical output power (total power at output
approximately 10 W)
â– Useful output voltages
– 12 V @0.65 A output (7.8 W, loose tolerance)
– 3.3 V @0.65 A output (2.145 W, tight tolerance)
â– Increased cost-effectiveness due to SMT construction (through-hole
connectors for
ruggedness)
> You are making this too complicated and needlessly expensive. One
> of my
> side projects right now is experimenting ...
>
> This is work in progress. I put the current snapshot at
> http://www.embedinc.com/temp/isol1.pdf. The transformer is from
> CoilCraft,
> and is meant for flyback operation for POE applications. It costs
> well
> under $1 in quantities.
From where? I can't find the "POE30..." from your snapshot at any of
the usual distributers (mouser, digikey, newark.) The cheapest "PoE"
transformers at mouser (from coiltronics or any other vendor) are in
that $4 range (q 100)...
> > You are making this too complicated and needlessly expensive.
There is a significant need for the demystifying of the inductor
aspects of switch mode power supplies. High quality cores suited to
the frequencies and power levels being discussed here are readily
available in the US (and elsewhere) at very modest cost and the
winding of inductors that meet typical needs is relatively simple and
well within the abilities of almost anyone who can spell EE (and of a
few who can't). This is far far far easier than eg what is involved in
winding AF or mains power transformers of yore.
I imagine that Gargoyle has much to say on this (I'll let you Gargoyle
that for me) but, if not, we have enough experience amongst list
members as to be able to provide useful design guidelines and a few
standard recipes.
'William Chops" Westfield ' <RemoveMEwestfwTakeThisOuTmac.com wrote:
>> The transformer is from CoilCraft,
>> and is meant for flyback operation for POE applications. It costs
>> well under $1 in quantities.
>
> From where?
> I can't find the "POE30..." from your snapshot at any of
> the usual distributers (mouser, digikey, newark.) The cheapest "PoE"
> transformers at mouser (from coiltronics or any other vendor) are in
> that $4 range (q 100)...
Be careful with specifying just "POE transformer". That could mean a signal
path transformer with center taps on the external side to pick off one of
the possible POE supplies. What I'm talking about is a power transformer
intended to make a isolated switching power supply with the primary
receiving about 48V switched.
>... The NSHUT line is driven low by the opto when the
> output is above its regulation threshold. This shuts down the PWM output
> automatically without firmware intervention.
You can use NSHUT to drive Q2 directly - POSSIBLY even using the opto output
transistor in place of Q2 ! :-). The processor doesn't have to know -
although you could provide it with the signal "for advice" if it was useful
in the larger scheme of things.
The principle works well and can produce results which are better than may
be expected.
I did something extremely similar* some years ago - even extending to the
LDO with one Vbe drop which I used to drive the SMPS feedback.
* "Similar" relates to the principle - the hardware used a packet of CD40106
gates for everything digital :-). I guess in retrospect that I expected the
drive clamp to work on whole cycles when regulation was reached BUT what
actually happened was that it adjusted the length of the drive pulses to
maintain very smooth cycle by cycle regulation. Intuitive in hindsight but a
pleasant discovery at the time.
The end result was very effective - there are an unknown number of them 'out
there'. Possibly only thousands, but possibly millions, as it was done in a
design implemented for a client by one of the major Taiwanese exercise
machine console makers who produce consoles for a wide range of products.
As it solved a problem shared by many other products it may [tm] have
'migrated" beyond my client's products. One of these days I'll have to tear
an unrelated product apart and see if they are using the design :-). (There
are unlikely to be many other CMOS inverted based SMPS's in Taiwanese
products :-) ). It was used to provide a stable local supply for the logic
core in the face of widely varying battery voltage** and sudden very large
and noisy electromechanical battery loads. AFAIR the CD40106 smps and 2 x
LDOs quiescent current loads and some logic idled "in regulation" at
slightly under 100 microamps. (** 4 x C cells and some logic rated at
nominal 5 volts. Not a marvellous design initially).
> I did something extremely similar* some years ago - even extending to the=
LDO with one Vbe drop which I used to drive the SMPS feedback.
> * "Similar" relates to the principle - the hardware used a packet of CD40=
106 gates for everything digital :-). I guess in retrospect that I expected=
the drive clamp to work on whole cycles when regulation was reached BUT wh=
at actually happened was that it adjusted the length of the drive pulses to=
maintain very smooth cycle by cycle regulation. Intuitive in hindsight but=
a pleasant discovery at the time.
> =A0 ... It was used to provide a stable local supply for the logic core =
in the face of widely varying battery voltage** =A0and sudden very large an=
d noisy electromechanical battery loads. AFAIR the CD40106 smps and 2 x LDO=
s quiescent current loads and some logic idled "in regulation" at slightly =
under 100 microamps. (** 4 x C cells and some logic rated at nominal 5 volt=
s. Not a marvellous design initially).
________________________
Here we are. Seemed worth digging up.
Y' don't buy no ugly SMPS!
I hacked this out of the middle of a jpg with other irrelevant stuff
in it and cut and pasted it into left to right alignment using
Irfanview so the result is even less beautiful than the ugly original.
As can be seem, the circuit is identical to Olins apart from not using
a PIC, a POE transformer, output isolation, optocoupler, or almost
anything else and using 2 CMOS Schmitt inverters instead.
:-)
Despite all that it is remarkably similar to Olin's version.
LH inverter is an oscillator with 2 x R + diode providing asymmetric mark/s=
pace.
RH inverter (RHI) drives transistor to provide drive to L21 which
"rings" to > Vbattery when Q21 is turned off.
When voltage on C22 at input of LDO regulator rises a Vbe drop =3Dabout
0.5 Volt above regulator output then Q22 starts to turn on.
If Q22 is turned partially on it will pull input of RHI high thereby
stopping drive to Q21.
The squeamish may wish to put a resistor betwixt Q22c and RHI pin 3.
If Q22 was turn hard on the drive would cease suddenly until C22
voltage fell far enough to turn Q22 off.
In practice Q22 turns on enough to be "modulated" by rising V_C22 to
reduce the effective length of the drive pulses to Q21. Look at it too
too hard and the logic begins to fail, but in practice this is what
happens.
The arrangement with Q22 and IC22 means that the LDO regulator is
always supplied with a voltage about 0.5 volt above its output. As
long as the LDO does not drop out with this delta-V it works well. I
show an LM2950, which is good enough for this, but the final version
used a superb and approx zero cost Taiwanese sourced LDO.
Frequency, mark space ratio and inductor values can be arranged to
suit. This would work well enough at milliWatts to Watts.
The 40106 allows very low Iq with no load.
I left R26 in, although its purpose is obscure - presumably it snubbed
a slight transient. Inductor value is high, reflecting low power level
here.
Note that the oscillator is loaded by the modulation but not stopped
and that right hand inverter only needs a sniff of signal SO this
system could 'borrow" a clock from anywhere available and us only a
single inverter. OR replace RH inverter with a MOSFET and pullup and
get a discrete parts add on smps.
1 x MOSFET, 2 x bipolar, L, C, LDO.
Bob Blick may (should ! :-) ) now present a 3 transistor SEPIC
equivalent to round out the offerings.
Russell McMahon wrote:
> You can use NSHUT to drive Q2 directly - POSSIBLY even using the opto
> output transistor in place of Q2 ! :-).
Yes, in theory the right signals would be in the right place. However, the
opto is kept just on either side of its trip point all the time. This does
not lend itself well to a low impedence drive. The processor performs a
useful service in converting this rather slowly changing line into a nice
digital signal. It is also tolerant of that signal bouncing a little due to
coupled noise, although I don't actually see much evidence of it bouncing on
the scope. The PWM generator is shut down immediately when NSHUT goes low,
but is not re-enabled until the start of the next regular PWM period.
The processor also produces nice strong and sharp edges on PULSE and NPULSE.
These keep the main switch transistor Q1 nicely full on or full off most of
the time. The purpose of Q2 is to speed up the turnoff of Q1, which would
otherwise take a while since it's deliberately driven into saturation when
on.
So while I think your idea could be made to work, especially at low power
levels where lower efficiency of Q1 can be tolerated, my scheme has some
benefits once you've already decided the PIC with its PWM module and
shutdown input will be there.
> The principle works well and can produce results which are better
> than may be expected.
That may be particularly applicable in a extra low cost design where the PIC
is replaced by a simple oscillator. The bias winding of the transformer
could probably be used to make a oscillator with few parts. Basically the
oscillator just runs and periodically pusles the primary. The feedback via
the opto kills the oscillations when the output is above its regulation
threshold. One version of my LED headlamp used a circuit like that,
although there was no opto envolved since it didn't need to be isolated.
cdb wrote:
> You have C3 as 1mF is that what you intend?
Yes. What isn't shown on the schematic is that it's specifically a low ESR
type. This one is a Panasonic FC series since I have those in stock. I've
been specifying the Panasonic FK series in this type of application in newer
designs. If this wasn't just a test board I probably would have paralleled
it with a 10uF ceramic.
What about it being 1mF surprises you? What did you expect it to be?
Russell McMahon wrote:
> As can be seem, the circuit is identical to Olins apart from not using
> a PIC, a POE transformer, output isolation, optocoupler, or almost
> anything else and using 2 CMOS Schmitt inverters instead.
> :-)
> Despite all that it is remarkably similar to Olin's version.
The one main difference between this circuit and the modified version of
mine you suggested is that the feedback goes into a digital gate. This
allows the feedback signal to be relatively high impedance and to change
slowly. The PIC performed this function in my circuit.
With this minor (but I think important) distinction, I have no problem with
your circuit. It should be extendable to driving a transformer instead of
just a inductor, and to using opto feedback. However, transformers are a
bit less forgiving about pulse width and duty cycle. Your circuit uses a
rather large inductor (330uH, could even be called "huge" in this context),
which allows for a lot of duty cycle and pulse width slop before bad things
happen. Unfortunately that's a lot harder to do with a transformer, at
least without overspecifying it.
Russell McMahon wrote:
> I think he expected it to be about 3 orders of magnitude smaller :-).
> ie - around 1 microFarad.
A quick calculation shows that would be way too small. The transformer
primary can roughly do 300mA at 300uH. That means the energy of individual
pulses can be up to 300uH * 300mA**2 / 2 = 13.5uJ. At 5.7 volts, that's 1
amp for 2.4uS. The change on the capacitor would be 1A * 2.4uS / 1uF =
2.4V. Three orders of magnitude less is rather better.
>> I think he expected it to be about 3 orders of magnitude smaller :-).
>> ie - around 1 microFarad.
> A quick calculation shows that would be way too small.
Yes. I was commenting on probable expectation. ie when people write mF
they usually mean uF.
So when other people SEE mF they usually assume it is a typo (or
braino) or read it as uF without realising that they have done so. In
most cases there is a double error which corrects.
I've never got into the habit of using mF, even though it is the
arguably most correct nomenclature. Quite apart from the above risk,
my brain has never felt comfortable with the usage - but it's never
told me why.
On Tue, Apr 13, 2010 at 1:10 PM, Russell McMahon <EraseMEapptechnzgmail.com> wrote:
>>> I think he expected it to be about 3 orders of magnitude smaller :-).
>>> ie - around 1 microFarad.
>
>> A quick calculation shows that would be way too small.
>
> Yes. I was commenting on probable expectation. ie when people write mF
> they usually mean uF.
> So when other people SEE mF they usually assume it is a typo (or
> braino) or read it as uF without realising that they have done so. In
> most cases there is a double error which corrects.
>
> I've never got into the habit of using mF, even though it is the
> arguably most correct nomenclature. Quite apart from the above risk,
> my brain has never felt comfortable with the usage - but it's never
> told me why.
>
I have some old (hollow-state) gear with caps marked in "MF". Pretty
sure they aren't megafarads... I think it is the memory of this
non-standard usage that causes caps to be marked "10000 uF" rather
than "10 mF".
Russell McMahon wrote:
> I've never got into the habit of using mF, even though it is the
> arguably most correct nomenclature. Quite apart from the above risk,
> my brain has never felt comfortable with the usage - but it's never
> told me why.
I grew up with microfarads and picofarads, sometimes even
micro-micro-farads.
At some point I decided that was stupid and deliberately started using more
normal engineering units, as is standard for most other things. It's been
so many years now that millifarads and nanofarads feel perfectly normal, and
I have to stop and think when someone says 10000 picofarads or .01
microfarads.
I don't know why capacitance in Farads evolved differently from the normal
practise of adjusting the prefix so that there are 1 to 3 digits left of the
point, but it's silly and it's time to put a stop to it.
> I don't know why capacitance in Farads evolved differently from the normal
> practise of adjusting the prefix so that there are 1 to 3 digits left of the
> point, but it's silly and it's time to put a stop to it.
Maybe.
But first you have to re-educate everyone or shoot them.
The fact that micro range is common and mili less so means that people
frequently use m for micro.
Henry - milli and micro common. Mega not
Ohm. - milli and micro rare. Meg more used.
Volt - milli and micro common. MV mainly lightning
Farad - micro common. M and m less so.
metre milli common. Micrometre for technical use.
All except Farad have at least one common member in m or M range.
>It's been so many years now that millifarads and nanofarads feel
>perfectly normal, and
>I have to stop and think when someone says 10000 picofarads or .01
>microfarads.
>
>I don't know why capacitance in Farads evolved differently from the normal
>practise of adjusting the prefix so that there are 1 to 3 digits left of the
>point, but it's silly and it's time to put a stop to it.
I'd agree, except for one thing: a lot of old-timers still think uF
when they see mf of even MF. That's because people were careless in
the old days and used those terms interchangeably. Also often used
was mmf to represent picofarads. Although not correct, its easy to
understand how it happened. It simply was inconceivable that
capacitors could actually approach or exceed 1F in capacity. (pun not intended)
I've been using nF for at least three decades but still prefer to see
or write 1000uF instead of 1mf. I don't have any issues when I see
"0.01uF" but automatically think of it as 10nF.
I have to thank the magazine Elektor for my preference to the term
'nF" - its just something that made way too much sense when I saw it
for the first time. None of the American electronics magazines used
nF back in the '60s or '70s or even the 80's.
I still look carefully when I see something like "1mF" on a schematic
- a few seconds of analysis usually tells me if in fact the does mF
mean thousands of uF. I do still have some old schematics where mf =
microfarads.
Dwayne Reid wrote:
> I'd agree, except for one thing: a lot of old-timers still think uF
> when they see mf of even MF.
So how is it ever supposed to end with that attitude? If you keep writing
mF to mean microfarad, then you're creating the next generation of future
old timers. At some point you have to put a stop to it. That point was
about 10 to 20 years ago here.
Fortunately you can tell from context most of the time whether a capacitor
is measured in milli or micro Farads. When I see "22MF" on a schematic of
the 200V plate supply for a 6L6, I'm pretty sure it doesn't mean
millifarads. So while we have to tollerate the occasional mislabled ancient
schematic, we must label new schematics correctly. Anyone doing it wrong
should be publicly ridiculed, the modern internet equivalent of being tarred
and feathered. It's bad enough to see ".001uF" instead of "1nF", but "1mF"
to mean 10**-6 Farads is just plain wrong.
On Tue, 13 Apr 2010 17:18:11 -0400, "Olin Lathrop" said:
> Dwayne Reid wrote:
> > I'd agree, except for one thing: a lot of old-timers still think uF
> > when they see mf of even MF.
>
> So how is it ever supposed to end with that attitude? If you keep
> writing
> mF to mean microfarad, then you're creating the next generation of future
> old timers. At some point you have to put a stop to it. That point was
> about 10 to 20 years ago here.
>
> Fortunately you can tell from context most of the time whether a
> capacitor
> is measured in milli or micro Farads. When I see "22MF" on a schematic
> of
> the 200V plate supply for a 6L6, I'm pretty sure it doesn't mean
> millifarads. So while we have to tollerate the occasional mislabled
> ancient
If I saw a 200 volt supply for a 6L6 I'd say it needs more voltage :)
I guess in the interest of clarity, it helps to use units that cater to
the intended audience, and add extra explanation when there is
ambiguity.
When it comes to capacitors, I only know one person who would refer to
1000uF as 1mF. Personally I wait until terminology is popular before
using it, so for me, millifarads are out.
> wait until terminology is popular before using it, so for me,
> millifarads are out
I've gor a pretty old cap (30+ years ?) marked '1mF' which is a
0.001F / 1 millifarad / 1000uF capacitor. It's the only one I
recall seeing marked 'm' after repairing and stripping a lot of
boards. So at least one maker went against the flow but it
didn't catch on. uF caps around the same vintage are often
marked "MF"
At 03:18 PM 4/13/2010, Olin Lathrop wrote:
>Dwayne Reid wrote:
> > I'd agree, except for one thing: a lot of old-timers still think uF
> > when they see mf of even MF.
>
>So how is it ever supposed to end with that attitude? If you keep writing
>mF to mean microfarad, then you're creating the next generation of future
>old timers. At some point you have to put a stop to it. That point was
>about 10 to 20 years ago here.
I think that you are missing the point I was trying to make. At no
time was I suggesting that the ancient nomenclature I mentioned
should continue to be used. I was instead trying to point out why
your more correct nomenclature can cause confusion.
As a result, I tend to continue to write 1000uF or 2200uF where you
might write 1mf or 2.2mF or 2m2F or 2m2 (if using the European method
of placing the units where the decimal point goes).
I think that another decade to two will have to pass before mF
becomes used as commonly as thousands of uF is today.
>>> I'd agree, except for one thing: a lot of old-timers still think
>>> uF when they see mf of even MF.
>>
>> So how is it ever supposed to end with that attitude? If you keep
>> writing mF to mean microfarad, then you're creating the next
>> generation of future old timers.
I don't think anyone is advocating continuing to use "mF" of "MF" for
micro Farads. If we just avoid those "M" prefixes entirely, they
might be usable for the original purposes in another generation.
(although frankly, I'm never going to be very fond of a prefix that
changes a value by thee orders of magnitude if anyone mistypes the
case...)
> Dwayne Reid wrote:
> > I'd agree, except for one thing: a lot of old-timers still think uF
> > when they see mf of even MF.
>
> So how is it ever supposed to end with that attitude? If you keep writing
> mF to mean microfarad, then you're creating the next generation of future
> old timers. At some point you have to put a stop to it. That point was
> about 10 to 20 years ago here.
>
> Fortunately you can tell from context most of the time whether a capacitor
> is measured in milli or micro Farads. When I see "22MF" on a schematic of
> the 200V plate supply for a 6L6, I'm pretty sure it doesn't mean
> millifarads. So while we have to tollerate the occasional mislabled ancient
> schematic, we must label new schematics correctly. Anyone doing it wrong
> should be publicly ridiculed, the modern internet equivalent of being tarred
> and feathered. It's bad enough to see ".001uF" instead of "1nF", but "1mF"
> to mean 10**-6 Farads is just plain wrong.
>
Do you do the same for mass (weight) and distance and say/write 1Mg instead of
1000kg and 1Mm instead of 1000km? If I would put 1Mg in writing, I think that
everyone would think that I meant 1mg (depending of context, of course). I
assume you do use the SI units here (since that is the correct way).
Mass in particular is kind of peculiar in this case because in normal language
we have also stripped off the gram part and only say kilo, which is even more
confusing (kilo what?).
Another thing that leads to confusion regarding prefixing numbers with letters
is the micro prefix which is correctly written with a micro sign. I can't even
write it here and then I mostly use the 'u' letter instead since it is looking
the closest. I wonder what people that doesn't use roman/latin alphabet think
about this.
I think this and the fact that many prefixes starts with an 'm' letter
(milli,micro,mega) has brought on the confusion in the first place and I think
it is going to be very hard to change it at this point, which is why you are
never going to get everyone to use MF for 1000uF and are probably going to
cause more confusion than correction with this (which this whole thread is a
proof of).
Don't get me wrong, I am all for doing it the right way and commend you in
every way for trying. With Farads it is only the fact that the size difference
between 1mF and 1uF is so obvious, that you can do it all the way from design
to purchasing and production.
/Ruben
Currency is another case where ==============================
Ruben Jönsson
AB Liros Electronic
Box 9124, 200 39 Malmö, Sweden
TEL INT +46 40142078
FAX INT +46 40947388 EraseMErubenspamspamBeGonepp.sbbs.se
==============================
>> 1Mg instead of 1000kg...
>
>That would be 1MT, right? (Metric Ton.) :-)
Quantities in everyday use have the advantage over technical terms
because they've evolved with distinct names
A couple of examples that don't have common multipliers
second minute hour day week
ounce pound stone hundredweight ton
And one that does
gram kilogram tonne
Ignorance of the multiplier+unit system causes ambiguities like milli /
micro. How often do you see GMS or even GRMS for grams or L
or Ltrs for litre on consumer product packaging ? Mildly annoying
when everything else about the packaging spelling is fine
Dwayne Reid wrote:
> As a result, I tend to continue to write 1000uF or 2200uF
> ...
> I think that another decade to two will have to pass before mF
> becomes used as commonly as thousands of uF is today.
But how is that supposed to happen if lots of people like you don't make a
effort to use the more standard engineering notation (1-3 digits left of
point)? You want me to push for it, and you'll tag along when you think
it's safe?
'William Chops" Westfield ' <RemoveMEwestfwKILLspammac.com wrote:
> (although frankly, I'm never going to be very fond of a prefix that
> changes a value by thee orders of magnitude if anyone mistypes the
> case...)
PoE negotiation can be simplified to a few components. Although I
imagine that PoE chips are cheaper now than when I designed that
solution. I don't recall all the details, but it involved two zeners,
a transistor, and a few passives. I think the tradeoff was that it
only requested the full power - requesting power for smaller loads was
slightly more complicated.
> 'William Chops" Westfield ' <spamBeGonewestfwSTOPspamEraseMEmac.com wrote:
>> I for one would very much like to see a PoE (Power over Ethernet)
>> circuit(s) capable of supplying 0.5 to 2W isolated power.
>
> That is easily available. I fail to see the problem. For example,
> Coilcraft makes POE transformers for three different power ranges, the
> lowest being 3W if I remember right.
>
> No matter what you do with the switcher though, you still have to do the POE
> negotiation. There are chips from various vendors, including TI and
> National for that. One National chip I saw used recently included the
> swither controller and the POE negotiation in the same chip.
>
>
> ********************************************************************
> Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
> (978) 742-9014. Gold level PIC consultants since 2000.
Em 14/4/2010 10:10, M. Adam Davis escreveu:
> PoE negotiation can be simplified to a few components. Although I
> imagine that PoE chips are cheaper now than when I designed that
> solution. I don't recall all the details, but it involved two zeners,
> a transistor, and a few passives. I think the tradeoff was that it
> only requested the full power - requesting power for smaller loads was
> slightly more complicated.
>
> -Adam
>
The only problem of requesting full power (class 0, from 0.44W to
12.95W) is that the switch won't detect whether the equipment is
consuming more or less than it should (unless it consumes more than the
maximum 12.95W).
To me, the classification is just a nice protection feature, but not a must.
Regards,
Isaac
__________________________________________________
Fale com seus amigos de graça com o novo Yahoo! Messenger http://br.messenger.yahoo.com/
Isaac Marino Bavaresco wrote:
> The only problem of requesting full power (class 0, from 0.44W to
> 12.95W) is that the switch won't detect whether the equipment is
> consuming more or less than it should (unless it consumes more than
> the maximum 12.95W).
>
> To me, the classification is just a nice protection feature, but not
> a must.
The real advantage is not so much protection, but that the switch can budget
it's presumably limited power. If a switch knows a class 1 device is
plugged into a port, as apposed to a class 0, that's a few watts it knows it
can give to other ports.
At 06:17 AM 4/14/2010, Olin Lathrop wrote:
>Dwayne Reid wrote:
> > As a result, I tend to continue to write 1000uF or 2200uF
> > ...
> > I think that another decade to two will have to pass before mF
> > becomes used as commonly as thousands of uF is today.
>
>But how is that supposed to happen if lots of people like you don't make a
>effort to use the more standard engineering notation (1-3 digits left of
>point)? You want me to push for it, and you'll tag along when you think
>it's safe?
Nope - that's not how I see it.
Two things to think about: the old-timers are getting out of the
business (and dying off, unfortunately) - and - large size capacitors
occupying small volumes are slowly becoming more common. I'm saying
that I think that it will take another decade for both of those
things to happen. At that time, using terms like "mF" will be
commonly accepted.
In the meantime, I think that writing "1000 uF" is entirely
reasonable and correct.
When might I use the term "mF"? Probably if I'm dealing with really
large capacitors. Instead of writing just 10,000 uF or 25,000 uF -
I'd probably write it both ways as "25,000uF (25mF)".
:: t's bad enough to see ".001uF" instead of "1nF", but "1mF"
My error in the first place, I was expecting to see extra 0000's with
uF and the milli size didn't register as I wasn't expecting it. Lame
excuse maybe, but true.
When I first arrived in Australia I was waffling on about nF's and
would find peoples eyes glazed over while they tried to work out how
many K's of pF or fractions of nF. In fact I've found, apart from the
last 4 years, many people in the US weren't nF friendly, in fact one
hobby magazine edited my nF's to uF's deducting publishing money as
they went.
It's a case of what one is used to I suspect going at a mild topic
tangent - I once made a cake from a Norwegian magazine - they use
decilitres a lot, in the UK it's rarely if ever used, my mind refused
to cope - liquid coffee cake anyone? Point being, whilst mathmatically
it should be apparent, my mind doesn't always kick in at first glance.
> I have to thank the magazine Elektor for my preference to the term
> 'nF" - its just something that made way too much sense when I saw it
> for the first time. None of the American electronics magazines used
> nF back in the '60s or '70s or even the 80's.
OK, I put a copy again at http://www.embedinc.com/temp/isol1.pdf. This time
don't wait a month to download it. Stuff in the temp directory will
dissappear at unpredictable times without notice.
planet.eon.net>
