Desired Sinewave Generator IC

[DH1]

Hello,

there are many ways to build a sinewave generator however todate I
have not seen a med-low cost high spec. minimal IC design that surely
is now possible. The ICL038 etc. are large ICs with eccentric
intercoupled freq./duty cycle adjustment & the NTE864 version costs
$54 at Mouser, while high pin count expensive DDS chips exist that
require programming. Many low pin count square wave generators do
exist (e.g., the SOT23-5 LTC1799) and can be coupled with, e.g., 8th
order elliptical low pass switched capacitor filters for a two chip
solution. One interesting chip is the MSI MSLOSC SO-8 programmable DDS
sinewave generator that also requires an external clock (where to
purchase such a chip ?). Low pin count uC's with internal D/A's can
also be programmed, with RC or active filtering added, and optional
crystals.

What would be useful for many applications is a low cost low power
low distortion one resistor programmable sinewave generator in an
SC23-5 pin package with GND, VCC, R_freq and Out pins (similar in
concept to the LTC1799). The fifth pin could be used for an optional
gain setting resistor. An internal high precision capacitor completes
the timing reference, and frequency accuracy could be, e.g., R_freq
accuracy+0.1% (though even 1% would be useful). Instead of R-freq
current programming could be used with graceful transition from one
freq. to another for modulation etc.

A variation on this is an SPI programmable MEMs clock version for
very high freq. accuracy.

DH

 

[DH1]

I am looking for a very small outline package such as the SOT23-5
with as few external components as possible. However the XR2206 is
indeed a good quality solution in a larger format.

DH

 

[Jon Kirwan]

Would a SOT23-6 work? I know the PIC10F20x comes in those and the 200
version costs about 28 cents in ones, since I recently bought a few.
You'd have to program it. If memory serves, the internal oscillator
is of the 1% variety. The footprint I'm using is 0.100" x 0.145,
outside corner to outside corner of the copper pads. (I apologize,
but I didn't get enough from your original post to know what you need
-- just that you mentioned a low pincount micro.)

Jon

 

[Michael Black]

Hello,

there are many ways to build a sinewave generator however todate I
have not seen a med-low cost high spec. minimal IC design that surely
is now possible. The ICL038 etc. are large ICs with eccentric
intercoupled freq./duty cycle adjustment & the NTE864 version costs
$54 at Mouser, while high pin count expensive DDS chips exist that
require programming. Many low pin count square wave generators do
exist (e.g., the SOT23-5 LTC1799) and can be coupled with, e.g., 8th
order elliptical low pass switched capacitor filters for a two chip
solution. One interesting chip is the MSI MSLOSC SO-8 programmable DDS
sinewave generator that also requires an external clock (where to
purchase such a chip ?). Low pin count uC's with internal D/A's can
also be programmed, with RC or active filtering added, and optional
crystals.

What would be useful for many applications is a low cost low power
low distortion one resistor programmable sinewave generator in an
SC23-5 pin package with GND, VCC, R_freq and Out pins (similar in
concept to the LTC1799). The fifth pin could be used for an optional
gain setting resistor. An internal high precision capacitor completes
the timing reference, and frequency accuracy could be, e.g., R_freq
accuracy+0.1% (though even 1% would be useful). Instead of R-freq
current programming could be used with graceful transition from one
freq. to another for modulation etc.

A variation on this is an SPI programmable MEMs clock version for
very high freq. accuracy.

DH

Well no.

ICs are designed and sold when there is either existing demand, or
a perceived demand can be shown.

Just because you want something doesn't mean there is real demand.
You haven't really even explained your need, you just want a simple
to use IC for reasons that haven't been explained.

There isn't a lot of need for a wide range audio sine wave generator.
No consumer device uses them, and most of the time when you need
an audio sine wave, a single frequency or a few specific frequencies
are all that are needed. Then you usually need a low distortion sinewave,
since a common use would be for distortion testing. The lowest
distortion schemes come easiest with those bulky components you don't
like. And realistically, by the time you have those LC or RC networks
to make a good sinewave, the cost of the needed active component (and
its size) add nothing to the project, so it's easier to switch between
multiple oscillators than switch the circuitry from frequency to
frequency.

Almost forty years ago, there was a conjunction where solid state
electronics allowed for putting a function generator in a single IC,
and yet where there was still a general demand for such a device. A
function generator is a general purpose device, so it covers a wide
range of frequency, and might have input for controlling it by voltage,
and has multiple output waveforms. Some outputs won't be perfect, which
is the nature of a general device. In this case, the sinewave output
is done by "forming" the triangle waveform with a matrix of diodes
and resistors. You get a sinewave, but it's not low distortion. The
various waveforms change as the duty cycle of the square wave output
changes because that's the nature of the design, you generate a triangle
by charging and discharging a capacitor, and the offshoot of that is
a square wave, but once you start tampering to change the triangle into
a ramp or change the duty cycle of the "square wave" output, that mangles
the other two outputs. But the circuitry is really easy, and that's
why such ICs became available.

Obviously at the time, they could be used in function generators,
eliminating much of the circuitry that had been needed previously. There
was also, though I have no idea how much the demand was in numbers,
audio synthesizers, where you wanted a voltage controlled oscillator
and multiple output waveforms.

One way or another, there was enough demand for the device, and any
further applications was icing on the cake. Back then, there wasn't
that much choice for audio frequency voltage controlled oscillators,
those function generator ICs like the 8038 and the 2206 that someone
else mentioned came more or less concurrent with the start of the
PLL synthesizer revolution. Those function generator ICs gave a lot
of flexibility, unlike the VCOs designed for PLLs.

But decades later, there still isn't much demand for a wide range
sinewave oscillator. Plus, other schemes have come along to fill
the need for oscillators. Analog synthesizers have pretty much
disappeared, digital techniques doing it a lot better. Anything
above a certain level would have a CPU anyway, so one might as
well use that to generate any needed waveforms.

If it was simple to generate a good sinewave, then there'd have
been such a device long ago. But it's not, so they have to consider
how much demand for it.

If there's little demand, then the cost goes up. It's too specialized
to be on the level of a 555. Chances are good that any time you
really need a audio frequency sinewave, the only loss of not having an IC
is the amount of space it takes up.

Michael

 

[DH1]

Well yes. I did not state that demand was high only that I desired it
and it would be useful for many applications. A low cost SOT23-5
design would minimize board space and passive requirements.
Applications include IR/Ultrasonic sensing/communication carrier
waves, high freq. power circuits, sine tone generation, etc. and the
list is likely almost as long as the number of contributors. I also
see other posts looking for a similar part, and there are many circuit
designs available attempting with different degrees of success to
provide minimal &/or low distortion sinewave generators.

If I need a single sinewave I will select a lower cost resistor
configured SPT23-5 part over an XR2206 everytime. The same would apply
for multiple sinewaves, or freq. &/or amplitude modulation designs.

Whether or not the part is currently economically viable I cannot be
certain, however based upon its merits I suspect that at the right
price it is. In the past I have found myself looking for something
that did not exist, yet did appear on the market a short time later
(SSDs are one example) and this can be a case of many people
recognizing a need and the required technological state of development
at the same time.

The prior development path that you outline is very interesting (and
I agree that the requirements for triangle/square & sine wave
generation are substantially different), however remarkable new
designs keep appearing on the market, such as programmable micro
crystal oscillators, digital pots, or a low cost 0.4uV offset 5MHz R2R
1.1mApower opamp in an SOIC8 format, as manufacturers push back the
boundaries. It may well soon be time to rethink the sinewave generator
and develope a med-high spec almost trivial to configure option(s) in
a minimal package, perhaps using PLL, MEMs, state machine/uC internal
control/stabilization, uPower techniques etc.

DH

 

[DH1]

It might indeed work, and I am looking at PIC and Silicon Labs uCs.
The six pins would limit an interface to an R/2R D/A, & no internal D/
A is included, so a filtered PWM approach would probably be best. The
filtered PWM approach produces a non-monotonic result, however if the
THD is acceptable this may well be the lowest outline design possible.

DH

 

[DH1]

It might indeed work, and I am looking at PIC and Silicon Labs uCs.
The six pins would limit an interface to an R/2R D/A, & no internal D/
A is included, so a filtered PWM approach would probably be best. The
filtered PWM approach produces a non-monotonic result, however if the
THD is acceptable this may well be the lowest outline design possible.

DH

 

[DH1]

Well, with all due respect, yes. I did not state that demand was high
only that I desired such a part and it would be useful for many
applications. A low cost SOT23-5 design would minimize board space and
passive requirements. Applications include IR/Ultrasonic sensing/
communication carrier waves, high freq. power circuits, sine tone
generation, etc. and the list is likely almost as long as the number
of contributors. I also see other posts looking for a similar part,
and there are many circuit designs available attempting with different
degrees of success to provide minimal &/or low distortion sinewave
generators.

If I need a single sinewave I will select a lower cost one or two
resistor configured SPT23-5 part over an XR2206 everytime. The same
would apply for multiple sinewaves, or freq. &/or amplitude modulation
designs.

Whether or not the part is currently economically viable I cannot be
certain, however based upon its merits I suspect that at the right
price it is. In the past I have found myself looking for something
that did not exist, yet did appear on the market a short time later
(SSDs are one example) and this can be a case of many people
recognizing a need and the required technological state of development
at the same time.

The prior development path that you outline is very interesting (and
I agree that the requirements for triangle/square & sine wave
generation are substantially different), however remarkable new
designs keep appearing on the market, such as programmable micro
crystal oscillators, digital pots, or a low cost 0.4uV offset 5MHz R2R
1.1mApower opamp in an SOIC8 format, as manufacturers push back the
boundaries. It may well soon be time to rethink the sinewave generator
and develope a med-high spec almost trivial to configure option(s) in
a minimal package, perhaps using PLL, MEMs, state machine/uC internal
control/stabilization, uPower techniques etc.

DH

 

[DH1]

Well, with all due respect, yes. I did not state that demand was high
only that I desired such a part and it would be useful for many
applications. A low cost SOT23-5 design would minimize board space and
passive requirements. Applications include IR/Ultrasonic sensing/
communication carrier waves, high freq. power circuits, sine tone
generation, etc. and the list is likely almost as long as the number
of contributors. I also see other posts looking for a similar part,
and there are many circuit designs available attempting with different
degrees of success to provide minimal &/or low distortion sinewave
generators.

If I need a single sinewave I will select a lower cost one or two
resistor configured SPT23-5 part over an XR2206 everytime. The same
would apply for multiple sinewaves, or freq. &/or amplitude modulation
designs.

Whether or not the part is currently economically viable I cannot be
certain, however based upon its merits I suspect that at the right
price it is. In the past I have found myself looking for something
that did not exist, yet did appear on the market a short time later
(SSDs are one example) and this can be a case of many people
recognizing a need and the required technological state of development
at the same time.

The prior development path that you outline is very interesting (and
I agree that the requirements for triangle/square & sine wave
generation are substantially different), however remarkable new
designs keep appearing on the market, such as programmable micro
crystal oscillators, digital pots, or a low cost 0.4uV offset 5MHz
1.1mApower opamp in an SOT23-5 format, as manufacturers push back the
boundaries. It may well soon be time to rethink the sinewave generator
and develope a med-high spec almost trivial to configure option(s) in
a minimal package, perhaps using PLL, MEMs, state machine/uC internal
control/stabilization, uPower techniques etc.

DH

 

[DH1]

Well, with all due respect, yes. I did not state that demand was high
only that I desired such a part and it would be useful for many
applications. A low cost SOT23-5 design would minimize board space and
passive requirements. Applications include IR/Ultrasonic sensing/
communication carrier waves, high freq. power circuits, sine tone
generation, etc. and the list is likely almost as long as the number
of contributors. I also see other posts looking for a similar part,
and there are many circuit designs available attempting with different
degrees of success to provide minimal &/or low distortion sinewave
generators.

If I need a single sinewave I will select a lower cost one or two
resistor configured SPT23-5 part over an XR2206 everytime. The same
would apply for multiple sinewaves, or freq. &/or amplitude modulation
designs.

Whether or not the part is currently economically viable I cannot be
certain, however based upon its merits I suspect that at the right
price it is. In the past I have found myself looking for something
that did not exist, yet did appear on the market a short time later
(SSDs are one example) and this can be a case of many people
recognizing a need and the required technological state of development
at the same time.

The prior development path that you outline is very interesting (and
I agree that the requirements for triangle/square & sine wave
generation are substantially different), however remarkable new
designs keep appearing on the market, such as programmable micro
crystal oscillators, digital pots, or a low cost 0.4uV offset 5MHz
1.1mApower opamp in an SOT23-5 format, as manufacturers push back the
boundaries. It may well soon be time to rethink the sinewave generator
and develop a med-high spec almost trivial to configure option(s) in a
minimal package, perhaps using PLL, MEMs, state machine/uC internal
control/stabilization, uPower techniques etc.

DH

 

[Jon Kirwan]

I picked up my <28 cent PIC10F200's in SOT23-6 at Newark. Just so you
know. Much cheaper than elsewhere. That's in 1's, by the way, though
you probably still want to order more or else get murdered with
shipping. SiLabs also has some tiny parts -- a 5-pin in some kind of
metal frame package that is TINY. But none so cheap!

Jon

 

[DH1]

A software delta-sigma DAC, with an output RC lowpass, is monotonic,
if you have the compute power. It trades noise for resolution.

John

Here on the other side of the pond the PIC10F200 is 0.64 EUR = $0.80
each, and from the same company as Newark (Farnell). The XR2206 is
also listed as no longer being stocked. However $0.80 is still good
value for low quantity projects. Farnell also have another related
company, CPC, with exceptional end of stock bargains, including R, C
and Zener 5x & 10x packs for as little as 0.01 EUR, and mains
transformers for as little as 0.40 EUR (I just bought a few, and two
will go back to back to make an isolation transformer).

Another possible sine generator option is the Cypress PSoC (I had
forgotten about these) and these are available at Farnell for about
2.00+ EUR.

By software delta-sigma do you mean a form of pulse delay modulation
and how do you get three states to the RC, +, 0 & - in order to avoid
a non-monotonic response ? Do you switch the port pin between +, high
impedance & 0V and send this to a mid rail biased capacitor ?

DH

 

[David L. Jones]

Well yes. I did not state that demand was high only that I desired it
and it would be useful for many applications. A low cost SOT23-5
design would minimize board space and passive requirements.
Applications include IR/Ultrasonic sensing/communication carrier
waves, high freq. power circuits, sine tone generation, etc. and the
list is likely almost as long as the number of contributors. I also
see other posts looking for a similar part, and there are many circuit
designs available attempting with different degrees of success to
provide minimal &/or low distortion sinewave generators.

If I need a single sinewave I will select a lower cost resistor
configured SPT23-5 part over an XR2206 everytime. The same would apply
for multiple sinewaves, or freq. &/or amplitude modulation designs.

Whether or not the part is currently economically viable I cannot be
certain, however based upon its merits I suspect that at the right
price it is. In the past I have found myself looking for something
that did not exist, yet did appear on the market a short time later
(SSDs are one example) and this can be a case of many people
recognizing a need and the required technological state of development
at the same time.

The prior development path that you outline is very interesting (and
I agree that the requirements for triangle/square & sine wave
generation are substantially different), however remarkable new
designs keep appearing on the market, such as programmable micro
crystal oscillators, digital pots, or a low cost 0.4uV offset 5MHz R2R
1.1mApower opamp in an SOIC8 format, as manufacturers push back the
boundaries. It may well soon be time to rethink the sinewave generator
and develope a med-high spec almost trivial to configure option(s) in
a minimal package, perhaps using PLL, MEMs, state machine/uC internal
control/stabilization, uPower techniques etc.

It gets my vote. I seem to be always in need of just such a simple sine wave
generator, the last time being last week when all I wanted was a 1kHz 1V
sine wave as a test signal on a board. A few % distortion would have been
fine.
It's always amazed me that a simple SOT-23-5 or SO8 device like that has not
been produced.
If you build it, they will come...

Dave.

 

[David L. Jones]

It might indeed work, and I am looking at PIC and Silicon Labs uCs.
The six pins would limit an interface to an R/2R D/A, & no internal D/
A is included, so a filtered PWM approach would probably be best. The
filtered PWM approach produces a non-monotonic result, however if the
THD is acceptable this may well be the lowest outline design possible.

How about the P89LPC micros?
The LPC904 has a DAC and is SO-8, but hard to get?
The LPC9102 also has a DAC in a TSSOP14 package for $1.65ea from Mouser.

Dave.

 

[[email protected]]

It gets my vote. I seem to be always in need of just such a simple sine wave
generator, the last time being last week when all I wanted was a 1kHz 1V
sine wave as a test signal on a board. A few % distortion would have been
fine.
It's always amazed me that a simple SOT-23-5 or SO8 device like that has not
been produced.
If you build it, they will come...

Dave.- Hide quoted text -

- Show quoted text -

Buy a used function generator off ebay and be done with it.

 

[DH1]

I had not considered the Philips/NXP series and Farnell/Newark stock
some of the range. They do appear to be good value.

DH

 

[DH1]

I understand. The feedback loop of the Sigma-Delta-Sigma takes care
of a-priori unknown waveforms and any variations in the following
filter and load (assuming you sample after the filter). For a fixed
sinewave and RC filter the output pulse sequence could be determined
beforehand and no feedback loop required as long as the DC offset was
set by one or more resistors. All other things being equal this should
allow for a faster output rate.

The audio Delta-Sigma usually has a high order low-pass filter to
remove ripple and some will pass through an RC filter. Since even with
a six pin uC solution there will be 4 port pins available, it makes
sense to consider a higher order Delta-Sigma approach using three (+/
0/-) or more output levels (4 pins allow for 7minus/0/7plus levels),
using one current setting resistor per pin, that will reduce filtering
requirements and ripple. The output current sequence is set beforehand
based upon the desired RC output error and error slope.

DH

 

[DH1]

I understand. The feedback loop of the Sigma-Delta-Sigma takes care
of a-priori unknown waveforms and any variations in the following
filter and load (assuming you sample after the filter). For a fixed
sinewave and RC filter the output pulse sequence could be determined
beforehand and no feedback loop required as long as the DC offset was
set by one or more resistors. All other things being equal this should
allow for a faster output rate.

The audio Delta-Sigma usually has a high order low-pass filter to
remove ripple and some will pass through an RC filter. Since even with
a six pin uC solution there will be 4 port pins available, it makes
sense to consider a higher order approach using three (+/0/-) or more
output levels (4 pins allow for 7minus/0/7plus levels), using one
current setting resistor per pin, that will reduce filtering
requirements and ripple. The output current sequence is set beforehand
based upon the desired RC output error, and slope error, and the
resistors do not even have to be in exactly a 1/2/4/8 ratio as any
close ratio can be taken into account when establishing the output
data.

DH

 

[David L. Jones]

It gets my vote. I seem to be always in need of just such a simple sine
wave
generator, the last time being last week when all I wanted was a 1kHz 1V
sine wave as a test signal on a board. A few % distortion would have been
fine.
It's always amazed me that a simple SOT-23-5 or SO8 device like that has
not
been produced.
If you build it, they will come...

Dave.- Hide quoted text -

- Show quoted text -

Buy a used function generator off ebay and be done with it.

LOL!
You obviusly don't understand what the OP is after, and what this thread is
about.
A function gen would be rather hard to solder onto a PCB, and would take up
considerable PCB realestate :->

Dave.

 

[whit3rd]

If it was simple to generate a good sinewave, then there'd have
been such a device long ago.  But it's not, so they have to consider
how much demand for it.

Not so much 'consider', as encourage someone to whine that there's
no demand, and allow managers to be scared off.

If it were REALLY an issue of 'no demand', the first such
oscillator-in-a-chip would be gathering dust on the shelf.
There's no such oscillator, because it IS a difficult task
to make a true, pure sinewave with a mass-producible
hardware base. The old HP 200 audio generator got its
second feedback loop (the amplitude control) from an
incandescent lamp (a component akin to 'ballast tubes').
Newer designs use the voltage-controlled-resistance
region of a selected-part jFET. In each case, it isn't
"standard" engineering practice, but a clever trick that
saves the design.

Making a tunable oscillator of the Wein bridge type requires
twin variable resistors (or capacitors); this is not easy to
do on a chip (unless you have very low expectations of
tuning range). It would be VERY useful (and would
obviate a lot of DDS hardware) to have a high
purity sinewave wide range VCO that could be locked to a
squarewave clock.

Digital sinewaves are expensive, analog triangle/sine
conversion is imprecise. Single-frequency high purity
generators are incompatible with logic control systems.

If there's little demand, then the cost goes up.

Let's disregard the economics of a design that doesn't
exist: the XR2206 DOES sell, and at a relatively high price
for a single chip. A better chip would probably sell, too,
and we won't know anything about the cost until we,
designers, design one.