Maker Pro
Maker Pro

Clean square waves using a 555 timer?

TedA

Sep 26, 2011
156
Joined
Sep 26, 2011
Messages
156
Xeno,

I have attempted to read the whole thread from the beginning, then address the more interesting issues outstanding. Please forgive me if I failed to take note of something I missed along the way.

Not all "555" timers are the same. Even bipolar parts vary among manufacturers and dates of manufacture. Even if the specifications are identical. CMOS parts are quite different from the bipolar original, and the CMOS specifications vary quite a bit among the various manufacturers. So it is often very useful to know the exact part number, manufacturer, and possibly even the date code for the part under discussion.

And, of course, it is a courtesy to provide a link to a data sheet for the version of the part you are using.

I found an old Signetics data sheet here: http://www.datasheetarchive.com/dl/Scans-005/Scans-00107395.pdf

And a more readable ST data sheet here: http://www.datasheetarchive.com/dl/Datasheets-EC1/DSAQ00222376.pdf

I hope I understand correctly that the "555" in question here is in fact a bipolar part, not CMOS.

It also appears that the part used by dorke to research the problem and post the nice Tek465 screen shots was a CMOS version. The bipolar parts will not produce an output = Vcc, even with no load. Guess that should be Vdd! I wonder how his part was marked.

I think that the tall narrow spike overshoot was probably due to spikes on Vcc caused by inductance in the Vcc wiring. Vcc at the device pins was actually spiking up above the nominal DC value when the output stage switched. As already pointed out, the bipolar NE555 draws a big current spike from Vcc when the output stage switches. Both output transistors are on at once for a brief time. When this spike ends, any inductance in series with the Vcc source will briefly pull the Vcc pin higher than normal.

I don't think any tantalum capacitors are really required for bypass use in this circuit. A fairly large ceramic cap backed-up by an aluminum electrolytic should be good enough. The plastic breadboard renders any really good capacitor beside the point, anyway.

( Should add that I have had fairly good luck with these breadboards, but they are not good for everything. Circuits using a 555, an LM324 or an LM339 seem to do well enough. Perhaps an MC1458 or uA741. 7400 or 74LS00 logic is usually OK. Fast stuff, very low level audio, or RF circuits may produce very unexpected results.)

Once the breadboard circuit construction was cleaned up a bit, the remaining overshoot may be due to transistor storage time effects. The bipolar 555 is made using an analog process. The transistors are not optimized for fast switching. Storage times may be in the hundreds of nanoseconds.

Here's that screen shot again, showing the 555 output pin transition from low to high.

upload_2016-3-13_21-47-59-png.25585


It appears that the overshoot seen above reaches very close to the 9VDC Vcc, and lasts for 300ns to 350ns.

I think this overshoot may be due to the output transistor storage time.

This transistor (Q22) gets lots of base drive as the output pin moves high. Base drive is applied before the pull-down transistor, Q24, begins to turn off. Maximum base drive current continues to be applied to Q22 as the output pin rises. Once the pin reaches about 1.5V below Vcc, the base drive current falls rapidly, but the stored charge keeps Q22 turned-on for a short time. With a light load on the output pin, Q24 will pull the output close to Vcc until the Q24 storage time interval is over.

After the end of the Q22 storage time, Q22 falls out of saturation, the output pin then falls to the steady state value, about 1.5 volts below Vcc. Around this voltage level, the base drive for the transistor is restored and holds it there.

Again, these are quite slow transistors. They turn on more quickly than they turn off. So when the output is switching, lots of current is flowing through them between Vcc and GND, even before the output pin begins to move.

Using the output pin to drive the timing resistor is perfectly OK, as long as the output pin's limitations are taken into account.

If you look at the more complete data sheets, you will see that the output voltage versus current characteristics are not described by simple straight lines. And neither the high or low output voltage is right against the rail. And these voltages depend on the load on the output.

So timing that is dependent on the output pin voltage may not be as precise as you would like. But there is no inherent reason not to use the output pin in the timing circuit. Sometimes a pull-up resistor can help this work better.

It is entirely likely that most SPICE models fail to accurately model the behavior of the output pin.

As already suggested, a CMOS "555" may be a better choice for you. The CMOS parts generally have more symmetrical output pin characteristics, and will pull light loads very close to both supply rails.

Using the CMOS part, you can produce a nominal 50% squarewave output using just one resistor and one capacitor. No need for diodes, pull up resistors or other extra stuff.

Of course, due to device tolerances, the exact % will vary from device to device, and to some extent with temperature. And a heavy load on the output pin will change the timing.

When considering CMOS alternatives, some study of the data sheets is a very good idea. Specifications vary from one variety to another. Many have much lower maximum power supply voltage ratings than the original's 18V. Often the output currents are rated much lower, also.

Ted
 

Xeno Xenox

Mar 13, 2016
50
Joined
Mar 13, 2016
Messages
50
Xeno,

I have attempted to read the whole thread from the beginning, then address the more interesting issues outstanding. Please forgive me if I failed to take note of something I missed along the way.

Not all "555" timers are the same. Even bipolar parts vary among manufacturers and dates of manufacture. Even if the specifications are identical. CMOS parts are quite different from the bipolar original, and the CMOS specifications vary quite a bit among the various manufacturers. So it is often very useful to know the exact part number, manufacturer, and possibly even the date code for the part under discussion.

And, of course, it is a courtesy to provide a link to a data sheet for the version of the part you are using.

I found an old Signetics data sheet here: http://www.datasheetarchive.com/dl/Scans-005/Scans-00107395.pdf

And a more readable ST data sheet here: http://www.datasheetarchive.com/dl/Datasheets-EC1/DSAQ00222376.pdf

I hope I understand correctly that the "555" in question here is in fact a bipolar part, not CMOS.

It also appears that the part used by dorke to research the problem and post the nice Tek465 screen shots was a CMOS version. The bipolar parts will not produce an output = Vcc, even with no load. Guess that should be Vdd! I wonder how his part was marked.

I think that the tall narrow spike overshoot was probably due to spikes on Vcc caused by inductance in the Vcc wiring. Vcc at the device pins was actually spiking up above the nominal DC value when the output stage switched. As already pointed out, the bipolar NE555 draws a big current spike from Vcc when the output stage switches. Both output transistors are on at once for a brief time. When this spike ends, any inductance in series with the Vcc source will briefly pull the Vcc pin higher than normal.

I don't think any tantalum capacitors are really required for bypass use in this circuit. A fairly large ceramic cap backed-up by an aluminum electrolytic should be good enough. The plastic breadboard renders any really good capacitor beside the point, anyway.

( Should add that I have had fairly good luck with these breadboards, but they are not good for everything. Circuits using a 555, an LM324 or an LM339 seem to do well enough. Perhaps an MC1458 or uA741. 7400 or 74LS00 logic is usually OK. Fast stuff, very low level audio, or RF circuits may produce very unexpected results.)

Once the breadboard circuit construction was cleaned up a bit, the remaining overshoot may be due to transistor storage time effects. The bipolar 555 is made using an analog process. The transistors are not optimized for fast switching. Storage times may be in the hundreds of nanoseconds.

Here's that screen shot again, showing the 555 output pin transition from low to high.

upload_2016-3-13_21-47-59-png.25585


It appears that the overshoot seen above reaches very close to the 9VDC Vcc, and lasts for 300ns to 350ns.

I think this overshoot may be due to the output transistor storage time.

This transistor (Q22) gets lots of base drive as the output pin moves high. Base drive is applied before the pull-down transistor, Q24, begins to turn off. Maximum base drive current continues to be applied to Q22 as the output pin rises. Once the pin reaches about 1.5V below Vcc, the base drive current falls rapidly, but the stored charge keeps Q22 turned-on for a short time. With a light load on the output pin, Q24 will pull the output close to Vcc until the Q24 storage time interval is over.

After the end of the Q22 storage time, Q22 falls out of saturation, the output pin then falls to the steady state value, about 1.5 volts below Vcc. Around this voltage level, the base drive for the transistor is restored and holds it there.

Again, these are quite slow transistors. They turn on more quickly than they turn off. So when the output is switching, lots of current is flowing through them between Vcc and GND, even before the output pin begins to move.

Using the output pin to drive the timing resistor is perfectly OK, as long as the output pin's limitations are taken into account.

If you look at the more complete data sheets, you will see that the output voltage versus current characteristics are not described by simple straight lines. And neither the high or low output voltage is right against the rail. And these voltages depend on the load on the output.

So timing that is dependent on the output pin voltage may not be as precise as you would like. But there is no inherent reason not to use the output pin in the timing circuit. Sometimes a pull-up resistor can help this work better.

It is entirely likely that most SPICE models fail to accurately model the behavior of the output pin.

As already suggested, a CMOS "555" may be a better choice for you. The CMOS parts generally have more symmetrical output pin characteristics, and will pull light loads very close to both supply rails.

Using the CMOS part, you can produce a nominal 50% squarewave output using just one resistor and one capacitor. No need for diodes, pull up resistors or other extra stuff.

Of course, due to device tolerances, the exact % will vary from device to device, and to some extent with temperature. And a heavy load on the output pin will change the timing.

When considering CMOS alternatives, some study of the data sheets is a very good idea. Specifications vary from one variety to another. Many have much lower maximum power supply voltage ratings than the original's 18V. Often the output currents are rated much lower, also.

Ted

@TedA:

Thank you for your thoughtful and highly informative response!

I agree that the earliest sharp spike I saw well above Vcc was inductance affecting Vcc. I've verified this by removing and experimenting with various bypass capacitors.

Your theory of transistor storage delay being the cause of the much wider 300-350ns pulse to Vcc supports my intuition and everything I've experienced so far with this circuit. I'll find the appropriate datasheets and schematics for the particular bipolar 555s I'm using and work through your explanation.

At this stage, my main interest in this particular circuit is to better understand the bipolar 555 and develop a better understanding of its quirks. I have a dozen or so on hand and would like to put them to good use.

It's also good to know from yourself and others that CMOS 555s would produce cleaner output. I'll acquire some the next time I'm buying parts. I also saw a programmable 555 that might make for some interesting experiments.
 

dorke

Jun 20, 2015
2,342
Joined
Jun 20, 2015
Messages
2,342
The 555 I used isn't a CMOS one it is labeled NE555and is bipolar.
The manufacture logo isn't a famous known one.
 

CDRIVE

Hauling 10' pipe on a Trek Shift3
May 8, 2012
4,960
Joined
May 8, 2012
Messages
4,960
Hello Ted and welcome to EP.

It's been a long thread which makes it easy to miss the replies that previously addressed all of your points. So, unless I overlooked something, it's doubtful that anyone is going to dispute them;....except for one. I agree with Dork's commentary of why using the Output pin for the (RC) voltage source is less than desirable.

Chris
 

dorke

Jun 20, 2015
2,342
Joined
Jun 20, 2015
Messages
2,342
Sure,
I found this photo on the web,tried to enhance the logo.

555 logo.png
 

CDRIVE

Hauling 10' pipe on a Trek Shift3
May 8, 2012
4,960
Joined
May 8, 2012
Messages
4,960
I've accepted that translation to languages other than Chinese is not a Chinese priority but Logos too!!!!o_O

Chris
 

hevans1944

Hop - AC8NS
Jun 21, 2012
4,878
Joined
Jun 21, 2012
Messages
4,878
The guv'mint would open that puppy up and look at the die if someone tried to sell it to them as a "genuine TI" part... well, at least if it was being sold in substantial quantities for use by the military or NASA. At my previous employer we sold a machine that could remove that epoxy layer-by-layer, taking high-resolution pictures each step along the way, and then building a 3D model with all those thousands of pictures a few days later.

Uncle Sam paid to develop the first one (Robomet 3D), but metallurgists are allegedly the folks who use them. They are quite expensive, but the principal is simple: abrade using a rotating table covered with a diamond slurry, wash, rinse, photograph, and repeat... takes several minutes for each cycle, but the end result is pretty and aids in the visualization of grain boundaries and metallurgical "stuff" like that. I asked whether any of them were being used to "reverse engineer" integrated circuit dies, but all I got was a blank stare, or "Why would anyone want to do that?" Yeah, riiight.
 

Xeno Xenox

Mar 13, 2016
50
Joined
Mar 13, 2016
Messages
50
No ,it isn't TI, it is different.

The picture I shared appears to be the same picture that you cropped and painted over.

Although it could obviously be a Chinese counterfeit, the unmodified logo looks similar to the TI logo. It certainly isn't symmetric like your paint-over. It's shaped roughly like the state of Texas, albeit a bit narrower and smoother around the edges.
 

dorke

Jun 20, 2015
2,342
Joined
Jun 20, 2015
Messages
2,342
The picture I shared appears to be the same picture that you cropped and painted over.

Although it could obviously be a Chinese counterfeit, the unmodified logo looks similar to the TI logo. It certainly isn't symmetric like your paint-over. It's shaped roughly like the state of Texas, albeit a bit narrower and smoother around the edges.
My camera has lost it's zooming ability(Old Canon S5).
The logo looks like this:
555 logo.JPG
 

Xeno Xenox

Mar 13, 2016
50
Joined
Mar 13, 2016
Messages
50
I apologize for the delay in soldering up the circuit. My son discovered amateur radio and so I've been immersed in that for the past month. On the upside, I'm now a licensed ham! :)

This is the circuit I built, except that I added 10uF tantalum + 100nF ceramic bypass capactiors.

tim58a.gif


Here's my build:

upload_2016-5-15_15-48-33.png

upload_2016-5-15_15-49-0.png

And here's the probe connection (the grounding clip is attached directly to pin 1 in the back):

upload_2016-5-15_15-50-28.png

The color balance is different between these photos, but they are the same board.

Here's the falling edge of the waveform (with Vcc at 9.0V):

upload_2016-5-15_15-51-44.png

And here's the rising edge:

upload_2016-5-15_15-52-4.png

As you can see, the "hump" is still present on the rising edge, as it always has been.

After all these experiments, I think @TedA proposed the most plausible theory. We should expect the stable high level to be short of Vcc and it is. The temporary rise to Vcc (i.e., the "hump") is being caused by transistor storage delays within the NE555s I'm using.

Thank you everyone for your help and advice!
 

Attachments

  • upload_2016-5-15_15-50-3.png
    upload_2016-5-15_15-50-3.png
    325.6 KB · Views: 107
  • upload_2016-5-15_15-51-26.png
    upload_2016-5-15_15-51-26.png
    4 KB · Views: 99

MIL-Q

Aug 11, 2016
2
Joined
Aug 11, 2016
Messages
2
Xeno, I had the identical leading edge spike problems that you have described and found your posts while I was looking for a solution. I am using the 555 to generate a 10kHz sqr wave and driving a 10K resistive load.
I didn't care why the 555 has the spike, just how to get rid of it.
My fix: Put an RC filter/divider on the output. It totally tamed the 0.5 volt spike. Brought it down to 10mV.
The network was 1K from pin 3 to a 450pf ceramic cap grounded at the other end. Output for next stage taken at the resistor /cap junction of course. There are probably better combinations/solutions but it works good enough for me at the moment.
 

Xeno Xenox

Mar 13, 2016
50
Joined
Mar 13, 2016
Messages
50
Xeno, I had the identical leading edge spike problems that you have described and found your posts while I was looking for a solution. I am using the 555 to generate a 10kHz sqr wave and driving a 10K resistive load.
I didn't care why the 555 has the spike, just how to get rid of it.
My fix: Put an RC filter/divider on the output. It totally tamed the 0.5 volt spike. Brought it down to 10mV.
The network was 1K from pin 3 to a 450pf ceramic cap grounded at the other end. Output for next stage taken at the resistor /cap junction of course. There are probably better combinations/solutions but it works good enough for me at the moment.

@MIL-Q Thanks for letting me know! If I experience side effects in a practical circuit, I'll give your solution a try!
 

AnalogKid

Jun 10, 2015
2,884
Joined
Jun 10, 2015
Messages
2,884
My fix: Put an RC filter/divider on the output. It totally tamed the 0.5 volt spike. Brought it down to 10mV.
The network was 1K from pin 3 to a 450pf ceramic cap grounded at the other end. Output for next stage taken at the resistor /cap junction of course.
Did you happen to measure the output rise time before and after the fix?

ak
 

MIL-Q

Aug 11, 2016
2
Joined
Aug 11, 2016
Messages
2
Carefully measure and record;no. But observe closely;yes. The rise time remained virtually unaffected and vertical but the spike was gone and only a small, sharply rounded knee remained. I was happy and moved on to other challenges in the circuits that follow.
 
Top