From: miso on
On Feb 1, 9:00 pm, John Larkin
<jjlar...(a)highNOTlandTHIStechnologyPART.com> wrote:
> On Mon, 1 Feb 2010 20:43:39 -0800 (PST), "m...(a)sushi.com"
>
>
>
> <m...(a)sushi.com> wrote:
> >On Feb 1, 11:34 am, John Larkin
> ><jjlar...(a)highNOTlandTHIStechnologyPART.com> wrote:
> >> On Sun, 31 Jan 2010 16:30:19 -0800 (PST), MooseFET
>
> >> <kensm...(a)rahul.net> wrote:
> >> >On Jan 31, 3:03 pm, Jamie
> >> ><jamie_ka1lpa_not_valid_after_ka1l...(a)charter.net> wrote:
> >> >> MooseFET wrote:
> >> >> > On Jan 31, 7:29 am, Glen Walpert <nos...(a)null.void> wrote:
>
> >> >> >>On Sat, 30 Jan 2010 10:05:13 -0800, John Larkin wrote:
>
> >> >> >>>On Sat, 30 Jan 2010 09:58:57 -0600, mook Johnson <m...(a)mook.net> wrote:
>
> >> >If you want more efficient operation, you could do this:
>
> >> >Version 4
> >> >SHEET 1 880 680
> >> >WIRE 240 -80 64 -80
> >> >WIRE 240 -64 240 -80
> >> >WIRE -320 -48 -448 -48
> >> >WIRE 64 -48 64 -80
> >> >WIRE 64 -48 -320 -48
> >> >WIRE 96 -16 80 -16
> >> >WIRE -320 48 -320 -48
> >> >WIRE -320 64 -320 48
> >> >WIRE -160 128 -320 128
> >> >WIRE 80 128 80 -16
> >> >WIRE 80 128 64 128
> >> >WIRE 240 128 240 32
> >> >WIRE 272 128 240 128
> >> >WIRE 400 128 336 128
> >> >WIRE 720 128 400 128
> >> >WIRE 240 144 240 128
> >> >WIRE 720 160 720 128
> >> >WIRE 400 176 400 128
> >> >WIRE -320 192 -320 128
> >> >WIRE -160 192 -320 192
> >> >WIRE 144 192 64 192
> >> >WIRE -448 256 -448 -48
> >> >WIRE -416 256 -448 256
> >> >WIRE -160 256 -416 256
> >> >WIRE 96 256 64 256
> >> >WIRE 240 256 240 224
> >> >WIRE 240 256 96 256
> >> >WIRE 400 256 400 240
> >> >WIRE 400 256 240 256
> >> >WIRE 512 256 400 256
> >> >WIRE 720 256 720 224
> >> >WIRE 720 256 592 256
> >> >WIRE 96 288 96 256
> >> >WIRE -416 304 -416 256
> >> >WIRE 848 304 784 304
> >> >WIRE -320 320 -320 272
> >> >WIRE -192 320 -320 320
> >> >WIRE 144 320 144 192
> >> >WIRE 144 320 -192 320
> >> >WIRE -320 336 -320 320
> >> >WIRE -192 336 -192 320
> >> >WIRE 720 352 400 352
> >> >WIRE 144 368 144 320
> >> >WIRE 400 368 400 352
> >> >WIRE 400 368 144 368
> >> >WIRE -416 400 -416 384
> >> >WIRE -192 400 -320 400
> >> >FLAG 96 288 0
> >> >FLAG -416 400 0
> >> >FLAG -320 400 0
> >> >FLAG 848 304 0
> >> >SYMBOL Misc\\NE555 -48 160 R180
> >> >SYMATTR InstName U1
> >> >SYMBOL voltage -416 288 R0
> >> >WINDOW 123 0 0 Left 0
> >> >WINDOW 39 24 132 Left 0
> >> >SYMATTR SpiceLine Rser=0
> >> >SYMATTR InstName V1
> >> >SYMATTR Value 15
> >> >SYMBOL res -336 32 R0
> >> >WINDOW 3 -64 71 Left 0
> >> >SYMATTR Value 2000
> >> >SYMATTR InstName R2
> >> >SYMBOL cap -208 336 R0
> >> >SYMATTR InstName C2
> >> >SYMATTR Value 0.01?f
> >> >SYMBOL res -336 176 R0
> >> >WINDOW 3 -60 41 Left 0
> >> >SYMATTR Value 1500
> >> >SYMATTR InstName R1
> >> >SYMBOL ind 224 128 R0
> >> >SYMATTR InstName L1
> >> >SYMATTR Value 1mh
> >> >SYMBOL schottky 336 112 R90
> >> >WINDOW 0 0 32 VBottom 0
> >> >WINDOW 3 32 32 VTop 0
> >> >SYMATTR InstName D2
> >> >SYMATTR Value 1N5819
> >> >SYMATTR Description Diode
> >> >SYMATTR Type diode
> >> >SYMBOL cap 384 176 R0
> >> >SYMATTR InstName C1
> >> >SYMATTR Value .5?f
> >> >SYMBOL res 192 -32 R90
> >> >WINDOW 0 0 56 VBottom 0
> >> >WINDOW 3 32 56 VTop 0
> >> >SYMATTR InstName R3
> >> >SYMATTR Value 1k
> >> >SYMBOL pnp 176 32 M180
> >> >SYMATTR InstName Q1
> >> >SYMATTR Value BCW68F
> >> >SYMBOL zener 704 160 R0
> >> >SYMATTR InstName D1
> >> >SYMATTR Value DFLZ33
> >> >SYMATTR Description Diode
> >> >SYMATTR Type diode
> >> >SYMBOL npn 784 352 R180
> >> >SYMATTR InstName Q2
> >> >SYMATTR Value 2N3904
> >> >SYMBOL res 608 240 R90
> >> >WINDOW 0 0 56 VBottom 0
> >> >WINDOW 3 32 56 VTop 0
> >> >SYMATTR InstName R4
> >> >SYMATTR Value 10k
> >> >TEXT 416 408 Left 0 !.tran 0 0.010 0 1 startup
>
> >> Here's my inverter, sort of similar.
>
> >>ftp://jjlarkin.lmi.net/Inverter.jpg
>
> >> The thevenin impedance seen by the emitter determines loop gain. I
> >> needed the ESR of the tantalum in the output to get good loop
> >> stability; startup dV/dT is low enough that a tantalum appears safe
> >> here.
>
> >> Load regulation is surprisingly good, about a tenth of a volt from 80
> >> to 500 mA.
>
> >> John
>
> >I see a start up problem here. Look at the UVL of the LM5112. Won't
> >this circuit whack the inductor while starting up?
>
> It's designed to start up at about 70% duty cycle. When the output
> gets close to -12, the transistor turns on and backs off the duty
> cycle to whatever it needs, which should be a bit over 50%.
>
> John

I'm talking about the first pulse. The P-fet will be on because of the
UVL being designed with a N-fet in mind. You should be able to see
this on a storage scope.

I see a lot of people rolling their own converters and have read the
justifications, but seriously, controller chips have (or should have
if done well) every contingency in mind. The first pulse, the initial
ramp up (soft start), etc. Linear Tech, Maxim, etc parts have been
tied to the whipping post and flogged without mercy. If you sell
millions of a component, even problems with 1% of them is a total
disaster.
From: Joerg on
miso(a)sushi.com wrote:
> On Feb 1, 9:00 pm, John Larkin
> <jjlar...(a)highNOTlandTHIStechnologyPART.com> wrote:
>> On Mon, 1 Feb 2010 20:43:39 -0800 (PST), "m...(a)sushi.com"
>>
>>
>>
>> <m...(a)sushi.com> wrote:
>>> On Feb 1, 11:34 am, John Larkin
>>> <jjlar...(a)highNOTlandTHIStechnologyPART.com> wrote:
>>>> On Sun, 31 Jan 2010 16:30:19 -0800 (PST), MooseFET
>>>> <kensm...(a)rahul.net> wrote:
>>>>> On Jan 31, 3:03 pm, Jamie
>>>>> <jamie_ka1lpa_not_valid_after_ka1l...(a)charter.net> wrote:
>>>>>> MooseFET wrote:
>>>>>>> On Jan 31, 7:29 am, Glen Walpert <nos...(a)null.void> wrote:
>>>>>>>> On Sat, 30 Jan 2010 10:05:13 -0800, John Larkin wrote:
>>>>>>>>> On Sat, 30 Jan 2010 09:58:57 -0600, mook Johnson <m...(a)mook.net> wrote:
>>>>> If you want more efficient operation, you could do this:
>>>>> Version 4
>>>>> SHEET 1 880 680
>>>>> WIRE 240 -80 64 -80
>>>>> WIRE 240 -64 240 -80
>>>>> WIRE -320 -48 -448 -48
>>>>> WIRE 64 -48 64 -80
>>>>> WIRE 64 -48 -320 -48
>>>>> WIRE 96 -16 80 -16
>>>>> WIRE -320 48 -320 -48
>>>>> WIRE -320 64 -320 48
>>>>> WIRE -160 128 -320 128
>>>>> WIRE 80 128 80 -16
>>>>> WIRE 80 128 64 128
>>>>> WIRE 240 128 240 32
>>>>> WIRE 272 128 240 128
>>>>> WIRE 400 128 336 128
>>>>> WIRE 720 128 400 128
>>>>> WIRE 240 144 240 128
>>>>> WIRE 720 160 720 128
>>>>> WIRE 400 176 400 128
>>>>> WIRE -320 192 -320 128
>>>>> WIRE -160 192 -320 192
>>>>> WIRE 144 192 64 192
>>>>> WIRE -448 256 -448 -48
>>>>> WIRE -416 256 -448 256
>>>>> WIRE -160 256 -416 256
>>>>> WIRE 96 256 64 256
>>>>> WIRE 240 256 240 224
>>>>> WIRE 240 256 96 256
>>>>> WIRE 400 256 400 240
>>>>> WIRE 400 256 240 256
>>>>> WIRE 512 256 400 256
>>>>> WIRE 720 256 720 224
>>>>> WIRE 720 256 592 256
>>>>> WIRE 96 288 96 256
>>>>> WIRE -416 304 -416 256
>>>>> WIRE 848 304 784 304
>>>>> WIRE -320 320 -320 272
>>>>> WIRE -192 320 -320 320
>>>>> WIRE 144 320 144 192
>>>>> WIRE 144 320 -192 320
>>>>> WIRE -320 336 -320 320
>>>>> WIRE -192 336 -192 320
>>>>> WIRE 720 352 400 352
>>>>> WIRE 144 368 144 320
>>>>> WIRE 400 368 400 352
>>>>> WIRE 400 368 144 368
>>>>> WIRE -416 400 -416 384
>>>>> WIRE -192 400 -320 400
>>>>> FLAG 96 288 0
>>>>> FLAG -416 400 0
>>>>> FLAG -320 400 0
>>>>> FLAG 848 304 0
>>>>> SYMBOL Misc\\NE555 -48 160 R180
>>>>> SYMATTR InstName U1
>>>>> SYMBOL voltage -416 288 R0
>>>>> WINDOW 123 0 0 Left 0
>>>>> WINDOW 39 24 132 Left 0
>>>>> SYMATTR SpiceLine Rser=0
>>>>> SYMATTR InstName V1
>>>>> SYMATTR Value 15
>>>>> SYMBOL res -336 32 R0
>>>>> WINDOW 3 -64 71 Left 0
>>>>> SYMATTR Value 2000
>>>>> SYMATTR InstName R2
>>>>> SYMBOL cap -208 336 R0
>>>>> SYMATTR InstName C2
>>>>> SYMATTR Value 0.01?f
>>>>> SYMBOL res -336 176 R0
>>>>> WINDOW 3 -60 41 Left 0
>>>>> SYMATTR Value 1500
>>>>> SYMATTR InstName R1
>>>>> SYMBOL ind 224 128 R0
>>>>> SYMATTR InstName L1
>>>>> SYMATTR Value 1mh
>>>>> SYMBOL schottky 336 112 R90
>>>>> WINDOW 0 0 32 VBottom 0
>>>>> WINDOW 3 32 32 VTop 0
>>>>> SYMATTR InstName D2
>>>>> SYMATTR Value 1N5819
>>>>> SYMATTR Description Diode
>>>>> SYMATTR Type diode
>>>>> SYMBOL cap 384 176 R0
>>>>> SYMATTR InstName C1
>>>>> SYMATTR Value .5?f
>>>>> SYMBOL res 192 -32 R90
>>>>> WINDOW 0 0 56 VBottom 0
>>>>> WINDOW 3 32 56 VTop 0
>>>>> SYMATTR InstName R3
>>>>> SYMATTR Value 1k
>>>>> SYMBOL pnp 176 32 M180
>>>>> SYMATTR InstName Q1
>>>>> SYMATTR Value BCW68F
>>>>> SYMBOL zener 704 160 R0
>>>>> SYMATTR InstName D1
>>>>> SYMATTR Value DFLZ33
>>>>> SYMATTR Description Diode
>>>>> SYMATTR Type diode
>>>>> SYMBOL npn 784 352 R180
>>>>> SYMATTR InstName Q2
>>>>> SYMATTR Value 2N3904
>>>>> SYMBOL res 608 240 R90
>>>>> WINDOW 0 0 56 VBottom 0
>>>>> WINDOW 3 32 56 VTop 0
>>>>> SYMATTR InstName R4
>>>>> SYMATTR Value 10k
>>>>> TEXT 416 408 Left 0 !.tran 0 0.010 0 1 startup
>>>> Here's my inverter, sort of similar.
>>>> ftp://jjlarkin.lmi.net/Inverter.jpg
>>>> The thevenin impedance seen by the emitter determines loop gain. I
>>>> needed the ESR of the tantalum in the output to get good loop
>>>> stability; startup dV/dT is low enough that a tantalum appears safe
>>>> here.
>>>> Load regulation is surprisingly good, about a tenth of a volt from 80
>>>> to 500 mA.
>>>> John
>>> I see a start up problem here. Look at the UVL of the LM5112. Won't
>>> this circuit whack the inductor while starting up?
>> It's designed to start up at about 70% duty cycle. When the output
>> gets close to -12, the transistor turns on and backs off the duty
>> cycle to whatever it needs, which should be a bit over 50%.
>>
>> John
>
> I'm talking about the first pulse. The P-fet will be on because of the
> UVL being designed with a N-fet in mind. You should be able to see
> this on a storage scope.
>
> I see a lot of people rolling their own converters and have read the
> justifications, but seriously, controller chips have (or should have
> if done well) every contingency in mind. ...


No, often they haven't. By far the biggest conceptual mistake is running
the internal VCC regulator at around 7.5V. Pretty much none of the FETs
above 200V Vds are guaranteed at less than 10V gate drive. Don't they
_read_ datasheets? What are they thinking? So you end up with some
external sauerkraut to correct that mistake with discrete parts. BTDT.
Great.

<banging head on table>

Then there's more, for example a lack of a 2nd FB input to
current-limit. So, more sauerkraut externally. At some point you might
as well roll the whole thing your self, and I often do.

I could go on. But the final clincher is the old $$$. A chip that costs
$2 or more in qties just doesn't fly in many designs. What the world
needs is a MC34063 in FET technology and running at up to a MHz. And a
similar TL494. Whoever brings that to market at reasonable pricing will
rake in tons of sales.

[...]

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
From: Tim Williams on
"Joerg" <invalid(a)invalid.invalid> wrote in message
news:7srk6kFk53U1(a)mid.individual.net...
> No, often they haven't. By far the biggest conceptual mistake is running
> the internal VCC regulator at around 7.5V. Pretty much none of the FETs
> above 200V Vds are guaranteed at less than 10V gate drive. Don't they
> _read_ datasheets? What are they thinking? So you end up with some
> external sauerkraut to correct that mistake with discrete parts. BTDT.
> Great.
>
> <banging head on table>

I was looking at a UCC something PFC controller the other day. It was
designed for 48-96V input service, for telecom supplies. Okay, fine. Maybe
it's still useful at 88-265V, just a few resistors away. It had a 3.3V
internal reference: whatever, the UC3842 gets away with 1.25V. But, they
powered the output stage from Vref! That's just barely within the range of
an average MOS driver (usually a guaranteed TTL level input), and not enough
for any transistor directly, even if it had a beefy output stage (which it
didn't, only ~10mA).

> Then there's more, for example a lack of a 2nd FB input to current-limit.
> So, more sauerkraut externally. At some point you might as well roll the
> whole thing your self, and I often do.

Why two current limits? Is that for, say, accurate output current limiting
(for a square output characteristic), plus inverter current limiting (via CT
or whatever)?

> I could go on. But the final clincher is the old $$$. A chip that costs $2
> or more in qties just doesn't fly in many designs. What the world needs is
> a MC34063 in FET technology and running at up to a MHz. And a similar
> TL494. Whoever brings that to market at reasonable pricing will rake in
> tons of sales.

Yeah, I was recently looking for forward controllers... now, there's
millions of flyback sorts of devices, and (usually with slope compensation)
these can be built into half wave forward converters too. But TL494 and
TL598 (totem pole output, cf. SG3524 and 3525) are the only push-pull
forward controllers I've seen. I was shocked to see there is well and
truely nothing new in this market, where the old, slow bipolar devices still
rule.

I would love to have one with beefy CMOS output (0.2 to 2A gate drive).
CMOS is more than capable of implementing the internal circuitry. A speed
upgrade besides and you've got a lovely IC.

What is LT doing? I know they've made lots of single-chip and HF sorts of
devices, but I never find anything from LT in searches like this.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms


From: Joerg on
Tim Williams wrote:
> "Joerg" <invalid(a)invalid.invalid> wrote in message
> news:7srk6kFk53U1(a)mid.individual.net...
>> No, often they haven't. By far the biggest conceptual mistake is running
>> the internal VCC regulator at around 7.5V. Pretty much none of the FETs
>> above 200V Vds are guaranteed at less than 10V gate drive. Don't they
>> _read_ datasheets? What are they thinking? So you end up with some
>> external sauerkraut to correct that mistake with discrete parts. BTDT.
>> Great.
>>
>> <banging head on table>
>
> I was looking at a UCC something PFC controller the other day. It was
> designed for 48-96V input service, for telecom supplies. Okay, fine. Maybe
> it's still useful at 88-265V, just a few resistors away. It had a 3.3V
> internal reference: whatever, the UC3842 gets away with 1.25V. But, they
> powered the output stage from Vref! That's just barely within the range of
> an average MOS driver (usually a guaranteed TTL level input), and not enough
> for any transistor directly, even if it had a beefy output stage (which it
> didn't, only ~10mA).
>

In medieval Europe they had a remedy for that: Dunking the people who
screwed up into a moat a couple of times, in front of an audience from
town. Had a very sobering effect :-)


>> Then there's more, for example a lack of a 2nd FB input to current-limit.
>> So, more sauerkraut externally. At some point you might as well roll the
>> whole thing your self, and I often do.
>
> Why two current limits? Is that for, say, accurate output current limiting
> (for a square output characteristic), plus inverter current limiting (via CT
> or whatever)?
>

Besides the current-mode control or a natural power limit you often need
to protect a load from over-current. The only way to do this with most
switchers is to have some external semiconductors and tug on the FB node
via a diode. Not pretty, had to do it many times.

The more common scenario in my case is that the client wants a steerable
current source but have it capped at xx volts so nothing gets fried if
the load opens up. Same problem, needs two FB inputs but there ain't
more than one.


>> I could go on. But the final clincher is the old $$$. A chip that costs $2
>> or more in qties just doesn't fly in many designs. What the world needs is
>> a MC34063 in FET technology and running at up to a MHz. And a similar
>> TL494. Whoever brings that to market at reasonable pricing will rake in
>> tons of sales.
>
> Yeah, I was recently looking for forward controllers... now, there's
> millions of flyback sorts of devices, and (usually with slope compensation)
> these can be built into half wave forward converters too. But TL494 and
> TL598 (totem pole output, cf. SG3524 and 3525) are the only push-pull
> forward controllers I've seen. I was shocked to see there is well and
> truely nothing new in this market, where the old, slow bipolar devices still
> rule.
>
> I would love to have one with beefy CMOS output (0.2 to 2A gate drive).
> CMOS is more than capable of implementing the internal circuitry. A speed
> upgrade besides and you've got a lovely IC.
>

Luckily there's the MIC4421/22 series but that always means two more
chips. Not much news in that domain :-(


> What is LT doing? I know they've made lots of single-chip and HF sorts of
> devices, but I never find anything from LT in searches like this.
>

My current favorite is the LT3757, sweet little chip, very tiny. Nice
thing is it can sense negative voltages. Plus they have LTSpice models
and that is the best thing since pivot irrigation. Less splatters,
transistor pieces and molten solder on the lab bench ;-)

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.
From: RST Engineering on
On Tue, 02 Feb 2010 14:42:42 -0800, Joerg <invalid(a)invalid.invalid>
wrote:


>
>In medieval Europe they had a remedy for that: Dunking the people who
>screwed up into a moat a couple of times, in front of an audience from
>town. Had a very sobering effect :-)
>

The original waterboarding?

Jim