Maker Pro
Maker Pro

Feedback for low frequency PWM regulator

N

nospam

Bitch, bitch, bitch, see updated driver at ABSE that controls Vrms to +/-3%
as input changes from 9VDC to 15VDC.

Good, a circuit which tries to do what is required.
--
 
H

HarryD

nospam said:
Good, a circuit which tries to do what is required.
--
"tries to do what is required" The OP wants to control RMS lamp voltage
driven by a discharging battery. The circuit shown can be trimmed to +/-1%
of a fixed RMS voltage for a battery input range of 15V to 9.0V. The +/-3%
above was just a fast cut.What else do you want?
Cheers,
Harry
 
J

JMini

By regulating the rms voltage across the lamp, you're not regulating
luminous intensity or power any more effectively than by regulating
the average voltage. For a resistor and pwm, these two voltage
quantities are the same.

Neither has a predictable relationship to power consumption in a load
with a positive temperature coefficient of resistance. PTC's with low
mass are strongly self-regulated for power. In the E^2 / R
relationship, as E increases, then so does R.

The use of the rms converter is simply complicating this circuit
unnecessarily. The use of PWM to reduce power loss in a control
element IS effective...you should be satisfied with this and
concentrate on simplifying the interface to a low-side switch and to
additionally reducing inrush current to values that will improve lamp
and switch operating life.

RL
As the input voltage falls, the PWM needs to increase to keep the RMS voltage
the same. I've tried regulating Average voltage. It doesn't work.
 
J

JMini

VFF will control average voltage across the lamp. Since the waveforms are
almost constant, the RMS voltage across the lamp is just a constant
factor. This is the first circuit you should try and then see what else is
needed. Tell me your requirements, input voltage range, switching
frequency, dimming control voltage, lamp current, lamp control voltage
range and any other info necessary. If you have parts already selected, I
will try to use them. A simple pdf design on
"alt.binaries.schematic.electronic" will be posted for your viewing
pleasure. Cheers,
Harry
I'm a Verizon customer. We just lost access to binariy groups. Is there a
resource on-line to get this pdf?
 
J

JMini

"tries to do what is required" The OP wants to control RMS lamp voltage
driven by a discharging battery. The circuit shown can be trimmed to +/-1%
of a fixed RMS voltage for a battery input range of 15V to 9.0V. The +/-3%
above was just a fast cut.What else do you want? Cheers,
Harry

I'm trying to find this schematic. Is there an web based tool to download
this schematic. F*cking Verizon.
 
L

legg

As the input voltage falls, the PWM needs to increase to keep the RMS voltage
the same. I've tried regulating Average voltage. It doesn't work.

This is one of the few applications where the average and rms voltage
are identical.If you averaged the voltage before it hit the resistive
load, there would be an issue. You're not doing this.

In both cases regulation is possible.

If you can't get the simpler arrangement working using an RC filter in
the feedback path, adding an rms converter will not improve the
situation.

RL
 
N

nospam

legg said:
This is one of the few applications where the average and rms voltage
are identical.If you averaged the voltage before it hit the resistive
load, there would be an issue. You're not doing this.

This is rubbish.

Average and RMS values are directly proportional to PWM duty cycle with the
same supply voltage.

The PWM duty cycles required to maintain a constant average or constant RMS
voltage with varying supply voltage are completely different.

20v with 25% duty and 10v with 50% duty give the same average voltage but a
2:1 difference in RMS voltage.
--
 
L

legg

On Mon, 14 Jul 2008 12:45:59 GMT, "JMini"


This is one of the few applications where the average and rms voltage
are identical.If you averaged the voltage before it hit the resistive
load, there would be an issue. You're not doing this.

In both cases regulation is possible.

If you can't get the simpler arrangement working using an RC filter in
the feedback path, adding an rms converter will not improve the
situation.

It's funny, considering the number of times I've gone through the
procedure of explaining RMS and average relationships, over the years,
to have to re-train myself on this matter. Sort of refreshing.

The formula Epk x rootD is so automatic that I seldom give it a second
thought. Just set me down with a pencil and paper, however, and I seem
to be able to prove to myself that the moon is made of green cheese.

Wonder why wasn't there wasn't more response, reacting to this
(temporary....?) lunacy? I suspect that spending too much time typoing
large databases must atrophy regular thought processes.

Sorry for the diversion.

RL
 
legg said:
It's funny, considering the number of times I've gone through the
procedure of explaining RMS and average relationships, over the years,
to have to re-train myself on this matter. Sort of refreshing.

The formula Epk x rootD is so automatic that I seldom give it a second
thought. Just set me down with a pencil and paper, however, and I seem
to be able to prove to myself that the moon is made of green cheese.

Wonder why wasn't there wasn't more response, reacting to this
(temporary....?) lunacy? I suspect that spending too much time typoing
large databases must atrophy regular thought processes.

Sorry for the diversion.

If this is the dumbest thing you have done in the last month, you
should break out the champagne and congratulate yourself. ;-)
 
J

Jamie

JMini said:
I'm a Verizon customer. We just lost access to binariy groups. Is there a
resource on-line to get this pdf?
Well now is the time to bail out of Verizon and go somewhere else with
out penalties. THey broke the contract of being a full internet
provider, so now its up to you!.



http://webpages.charter.net/jamie_5"
 
J

JMini

"tries to do what is required" The OP wants to control RMS lamp voltage
driven by a discharging battery. The circuit shown can be trimmed to +/-1%
of a fixed RMS voltage for a battery input range of 15V to 9.0V. The +/-3%
above was just a fast cut.What else do you want? Cheers,
Harry

This looks nice and simple.
What is the switching frequency? It looks like 47kHz is as low as this thing
can go. It looks like the UVLO would kick in at ~8.5V. Is there another IC
with a lower UVLO threshold? I'd like to just design a single circuit that
could operate from 6V to ~38-40V and just set the output voltage from
~0.6*Vin up to Vin. I would warn the user to avoid using battery voltages
that exceed desired RMS output by too much.
I could use an LDO between Vin and Vcc to limit Vcc voltage to something like
6V and adjust R5 to maintain voltage at the Comp pin to the desired range. A
Zener from Comp to GND would limit voltage for safety purposes. The output
(pin 6) I see is driving the gates of the power FET (Q1) , but also a P-ch
FET Q3 which grounds VR through R7 when Pin 6 goes low. What is the reason to
pull Vref low during Toff? Sorry if some of these questions seem naive, but
this is quite new to me.
 
J

JMini

This looks nice and simple.
What is the switching frequency? It looks like 47kHz is as low as this
thing can go. It looks like the UVLO would kick in at ~8.5V. Is there
another IC with a lower UVLO threshold? I'd like to just design a single
circuit that could operate from 6V to ~38-40V and just set the output
voltage from ~0.6*Vin up to Vin. I would warn the user to avoid using
battery voltages that exceed desired RMS output by too much.
I could use an LDO between Vin and Vcc to limit Vcc voltage to something
like 6V and adjust R5 to maintain voltage at the Comp pin to the desired
range. A Zener from Comp to GND would limit voltage for safety purposes.
The output (pin 6) I see is driving the gates of the power FET (Q1) , but
also a P-ch FET Q3 which grounds VR through R7 when Pin 6 goes low. What
is the reason to pull Vref low during Toff? Sorry if some of these
questions seem naive, but this is quite new to me.

Sorry for the reply to myself...
How about the UCC1803? Vcc-max = 12V
UVLO at 4.1V Pin compatible with the UC3842/A.
It has a lower Vcc-max, but if I use an LDO between the battery and Vcc, it
should be OK at battery voltages higher than 12V.
 
J

JMini

"tries to do what is required" The OP wants to control RMS lamp voltage
driven by a discharging battery. The circuit shown can be trimmed to +/-1%
of a fixed RMS voltage for a battery input range of 15V to 9.0V. The +/-3%
above was just a fast cut.What else do you want? Cheers,
Harry
Your design states a soft-start is on the "to-do" list.
If I substitute the UCC1803/3803, how would a softstart be implemented? I
need to start this thing slowly. 500mS is in the right range.
 
H

HarryD

JMini said:
Your design states a soft-start is on the "to-do" list.
If I substitute the UCC1803/3803, how would a softstart be implemented? I
need to start this thing slowly. 500mS is in the right range.

Many solutions, here is simplest; insert an 'Inrush Current Limiter"
(#MS320R536 at DK) in series with the load, 0.50 ohms cold and 25mR running.
Other more elegant solutions upon request.
If you are going to operate down to 4.5VDC, select a power MOSFET which
has low Ron @ 4.5Vgs. The operating frequency of 260Hz was stated on the
schematic.
Cheers,
Harry
 
J

JMini

Many solutions, here is simplest; insert an 'Inrush Current Limiter"
(#MS320R536 at DK) in series with the load, 0.50 ohms cold and 25mR
running. Other more elegant solutions upon request.
If you are going to operate down to 4.5VDC, select a power MOSFET which
has low Ron @ 4.5Vgs. The operating frequency of 260Hz was stated on the
schematic.
Cheers,
Harry
I thought perhaps it was a typo. Maybe 260 Kilohertz. But 260Hz is just fine.
The voltage in use will be no lower than 6V. For the greatest majority of
applications it will be over 10V. An FET driver like the TPS2819 can have a
Max input of 40V with an internal regulator to reduce voltage to 14V max to
the gate. Below 14V, the onboard regulator would just be in dropout. For
battery voltages lower than 10V I would select an FET like the IRLR7843. Good
Rds(on) at Vgs = 4.5V. The IRF2804 would be used for more powerful
applications. I've seen folks use the kind of Inrush Current Limiter you've
mentioned. Personally I'd rather do soft-starting on a time basis rather than
a current limit based on series resistance. Also, at 30mm x 8mm, it's about
the same size as the space I have for the entire circuit.
Can I request one of the "more elegant solutions"?
I'm still trying to wrap my brain around the workings of your schematic.

I see that with Rt=20K, the maximum duty cycle is limited to ~97%, would it
be possible to increase the size of Rt to 100K and reduce Ct to 68n? That
would keep the frequency roughly in the range you had (253Hz) Also, if the
UCC2803 is substituted the frequency is determined by 1/RC, But specs
recommend a Max Ct=1n and Rt=200K. That works out to 5kHz. Is a lower
frequency just not possible? Or is increasing Ct to 22n possible even though
it's way outside the recommended range?

I know this is a lot to throw at you, but you seem to have SPICE software and
the brains to use it.

Thanks, Harry.
 
H

HarryD

JMini said:
I thought perhaps it was a typo. Maybe 260 Kilohertz. But 260Hz is just
fine.
The voltage in use will be no lower than 6V. For the greatest majority of
applications it will be over 10V. An FET driver like the TPS2819 can have
a
Max input of 40V with an internal regulator to reduce voltage to 14V max
to
the gate. Below 14V, the onboard regulator would just be in dropout. For
battery voltages lower than 10V I would select an FET like the IRLR7843.
Good
Rds(on) at Vgs = 4.5V. The IRF2804 would be used for more powerful
applications. I've seen folks use the kind of Inrush Current Limiter
you've
mentioned. Personally I'd rather do soft-starting on a time basis rather
than
a current limit based on series resistance. Also, at 30mm x 8mm, it's
about
the same size as the space I have for the entire circuit.
Can I request one of the "more elegant solutions"?
I'm still trying to wrap my brain around the workings of your schematic.

I see that with Rt=20K, the maximum duty cycle is limited to ~97%, would
it
be possible to increase the size of Rt to 100K and reduce Ct to 68n? That
would keep the frequency roughly in the range you had (253Hz) Also, if the
UCC2803 is substituted the frequency is determined by 1/RC, But specs
recommend a Max Ct=1n and Rt=200K. That works out to 5kHz. Is a lower
frequency just not possible? Or is increasing Ct to 22n possible even
though
it's way outside the recommended range?

I know this is a lot to throw at you, but you seem to have SPICE software
and
the brains to use it.

Thanks, Harry.

You design like a TI apps. engineer, lots of silicon. Let's see a
schematic so we can remove parts.
The UCC2803 is a good choice. At your switching frequency, 250Hz, only
1.0mA is needed for supply current. So with input voltage from 30VDC to
50VDC, a 18K/0.25W resistor will suffice, no regulator is needed. 10K<
Rt<150K will work just fine, so Rt=120K and Ct =33nF will yield 250Hz and D
=~ 0.995. The internal soft start of 4mS will be of no help, you need
What happened to the other members of this NG? Do I have to do a complete
design here? This is turning into my day job with no benefits. JF could do
this with some 555s. Any more help will require good Tequila.
Cheers,
Harry
 
J

JMini

You design like a TI apps. engineer, lots of silicon. Let's see a
schematic so we can remove parts.
The UCC2803 is a good choice. At your switching frequency, 250Hz, only
1.0mA is needed for supply current. So with input voltage from 30VDC to
50VDC, a 18K/0.25W resistor will suffice, no regulator is needed. 10K<
Rt<150K will work just fine, so Rt=120K and Ct =33nF will yield 250Hz and
D =~ 0.995. The internal soft start of 4mS will be of no help, you need
What happened to the other members of this NG? Do I have to do a complete
design here? This is turning into my day job with no benefits. JF could do
this with some 555s. Any more help will require good Tequila. Cheers,
Harry
Ahh Harry, you're back. Just couldn't stand idly by and watch me wallow in
incompetence. Thus far I've just been Looking for uc3843 chips like the 2803
that have the lower voltage range I 've been looking for.
The ucc2803 has the lower UVLO that I need and you say it should be fine.
Great. All I have are hand drawn schematics at the moment. I was going to
order some components and build your design and start fiddling with it. For a
softstart, using linear regulators at least, you can produce a softstart but
holding down the Vref with a capacitor. But your design has the Vref switched
by the BSS84. I'd have to figure out what's going on there. From your design,
it looks like I can adjust output voltage by varying the values of the R5s
(46.4K and 13.3K)
 
H

HarryD

JMini said:
Ahh Harry, you're back. Just couldn't stand idly by and watch me wallow in
incompetence. Thus far I've just been Looking for uc3843 chips like the
2803
that have the lower voltage range I 've been looking for.
The ucc2803 has the lower UVLO that I need and you say it should be fine.
Great. All I have are hand drawn schematics at the moment. I was going to
order some components and build your design and start fiddling with it.
For a
softstart, using linear regulators at least, you can produce a softstart
but
holding down the Vref with a capacitor. But your design has the Vref
switched
by the BSS84. I'd have to figure out what's going on there. From your
design,
it looks like I can adjust output voltage by varying the values of the R5s
(46.4K and 13.3K)

Ok, ok, updated schematic posted on ABSE with the UCC2803 and soft start
added.
Q2 (BSS84) is not controlling VR, it is controlling Q3. That is how we
generate the voltage ramp for VFF control.

Cheers,
Harry
 
Top