555 Falling edge glitch

[pyrohaz]

Hi,

I'm using a TA7555 timer to produce a square wave (pretty standard!) at a relatively low frequency of 67kHz. The 555 timer is wired up as a schmitt trigger style oscillator (schematic included below).

Now just to point out. I am wiring the 555 timer up on a breadboard though my wiring is relatively neat and the waveform isn't glitching (I'll include photos too) but what I'm wondering is: Why does the falling edge of the timer contain a glitch at roughly Vcc/2?

The rising edge isn't particularly attractive either though I'm assuming this is because of breadboard capacitances and 555 timer output impedance. Obviously I'm not expecting simulation perfect outputs but I am wondering however why the output is so as it is?

The first picture is a close up of the falling edge. The second picture is the overall waveform and the final picture is the schematic!

Thanks,

[mij59]

Hi,

Could be caused by the weird output stage of the 555.
Try loading the output.

[SeanB]

C3 needs to have a big value capacitor across it, typically 470uF. The 555 timer has a very bad gotcha where it draws a large current pulse during switching, so the capacitor needs to be there as close to the device as possible to provide this large current pulse. On the breadboard place it between pins 1 and 8 in the holes right next to the chip.

[T3sl4co1l]

Note that 555 is not 7555, so your schematic isn't the same as what you've wired.

Mind a picture of the breadboard anyway?

Tim

[pyrohaz]

Hi,

Could be caused by the weird output stage of the 555.
Try loading the output.

I've added loading from 1k up to 33k. Below 10k, it doesn't seem to oscillate at all and at 33k, the frequency is drastically changed! Its only going to be driving the high impedance pin of a microcontroller anyway so is a load vital? Adding a 100k resistor reduces the total amplitude, along with stretching the frequency a bit too.

C3 needs to have a big value capacitor across it, typically 470uF. The 555 timer has a very bad gotcha where it draws a large current pulse during switching, so the capacitor needs to be there as close to the device as possible to provide this large current pulse. On the breadboard place it between pins 1 and 8 in the holes right next to the chip.

I didn't have any 470uF capacitors to hand so I've added a 2200uF capacitor. That doesn't seem to make any difference to the falling edge though it did decrease the rise time of the rising edge! I tried a 220uF capacitor and it made the rising edge slightly faster too.

Note that 555 is not 7555, so your schematic isn't the same as what you've wired.

Mind a picture of the breadboard anyway?

Tim

Hi Tim,

Yeh sorry about that, I quickly whipped up the schematic in LTSpice and don't have a model for the TA7555. As far as I remember, the LTSpice model is idealised anyway as the normal NE555 doesn't work down to this voltage IIRC.

The arduino on the board is doing nothing and the STM32F0 discovery board is just supplying power

[mij59]

Hi,

Could be caused by the weird output stage of the 555.
Try loading the output.

I've added loading from 1k up to 33k. Below 10k, it doesn't seem to oscillate at all and at 33k, the frequency is drastically changed! Its only going to be driving the high impedance pin of a microcontroller anyway so is a load vital? Adding a 100k resistor reduces the total amplitude, along with stretching the frequency a bit too.

This is very strange, what is the supply voltage of the timer ?

[Richard Head]

Start by connecting the ground lead of the scope probe to the negative rail. I know it's convenient to leave it off but when you work with high frequency edges you need a proper ground.

[pyrohaz]

Hi,

Could be caused by the weird output stage of the 555.
Try loading the output.

I've added loading from 1k up to 33k. Below 10k, it doesn't seem to oscillate at all and at 33k, the frequency is drastically changed! Its only going to be driving the high impedance pin of a microcontroller anyway so is a load vital? Adding a 100k resistor reduces the total amplitude, along with stretching the frequency a bit too.

This is very strange, what is the supply voltage of the timer ?

The timer is being ran at 3v which after further reading seems to be less than the minimum voltage of the TA7555! Argh, how silly of me, I thought the whole point of the 7xxx prefix was to denote low voltage versions but I'm obviously mistaken. Could this be one of the reasons why the output is so strange?

Start by connecting the ground lead of the scope probe to the negative rail. I know it's convenient to leave it off but when you work with high frequency edges you need a proper ground.

If you can see the wire in the bottom right of the breadboard photo, that is connected directly to the ground on my oscilloscope. Would it be more wise for me to ground the oscilloscope probe using a proper ground clip instead?

[Richard Head]

Would it be more wise for me to ground the oscilloscope probe using a proper ground clip instead?

Yes, it would be a better idea to ground the probe using the ground wire. If you don't then the coax cable doesn't behave as coax but just a shielded cable. Big difference.

[mij59]

The timer is being ran at 3v which after further reading seems to be less than the minimum voltage of the TA7555! Argh, how silly of me, I thought the whole point of the 7xxx prefix was to denote low voltage versions but I'm obviously mistaken. Could this be one of the reasons why the output is so strange?


Yes could be.

The min. power supply voltage is 5V, if you want to power it from 3V check the cmos version of the 555.

[T3sl4co1l]

I always thought 7555 *was* CMOS... 

Oh well, I never use the things anyway

Tim

[pyrohaz]

I always thought 7555 *was* CMOS... 

Oh well, I never use the things anyway

Tim

I'm glad I'm not alone here! It seems that Toshiba just don't do it that way. I generally find that without using an MCU, a 555 is the easiest way to create a square wave clock, care to enlighten me otherwise? 

Would it be more wise for me to ground the oscilloscope probe using a proper ground clip instead?

Yes, it would be a better idea to ground the probe using the ground wire. If you don't then the coax cable doesn't behave as coax but just a shielded cable. Big difference.

I'll keep that in mind for the future, thank you!

The timer is being ran at 3v which after further reading seems to be less than the minimum voltage of the TA7555! Argh, how silly of me, I thought the whole point of the 7xxx prefix was to denote low voltage versions but I'm obviously mistaken. Could this be one of the reasons why the output is so strange?


Yes could be.

The min. power supply voltage is 5V, if you want to power it from 3V check the cmos version of the 555.

I'll be buying some CMOS 555 timers soon. This time round, I'll read the datasheet in more depth before purchase!

Thank you all for your help.

[Christopher]

If you've got a big budget then the Linear Tech TimerBlox are quite nice,

PIC10F series 6 pin Sot23 are pretty neat for this kind of thing. But then you have the added complexity of added firmware

[T3sl4co1l]

I'm glad I'm not alone here! It seems that Toshiba just don't do it that way. I generally find that without using an MCU, a 555 is the easiest way to create a square wave clock, care to enlighten me otherwise? 

On the few occasions I need just a dumb constant square wave, well -- these days, there's those LT TimerBlox.. if you don't mind the extra buck.  Otherwise, I'll do a number of things; breadboarding, I might use a two transistor multivibrator, or a hysteresis comparator (same thing as a 555, but only needs one comparator -- albeit a pile of resistors around it).  The latter also generates a crude ramp, which leads into PWM (you can do (linear, constant frequency) PWM with a single 556, but I can do it with a single LM393).

By and large, I don't need constant frequency square waves; when it comes to switching supplies, most of mine look like either a two loop PWM system, or a variable astable with peak current mode control (something like a UC3842, with the oscillator supply strapped to the COMP pin -- so frequency is variable, keeping idle current down).

Tim

[pyrohaz]

And so the story can be solved! Today, my TLC555 timers arrived. Not only is the output waveform now symmetrical (no longer has that weird non-cmos 555 output stage), the waveform is pretty gorgeous if I say so my self! It turns out that TA7555 is obviously just not specced to that low voltage which is stated by the datasheet and a stupid error on my behalf!

I'm glad I'm not alone here! It seems that Toshiba just don't do it that way. I generally find that without using an MCU, a 555 is the easiest way to create a square wave clock, care to enlighten me otherwise? 

On the few occasions I need just a dumb constant square wave, well -- these days, there's those LT TimerBlox.. if you don't mind the extra buck.  Otherwise, I'll do a number of things; breadboarding, I might use a two transistor multivibrator, or a hysteresis comparator (same thing as a 555, but only needs one comparator -- albeit a pile of resistors around it).  The latter also generates a crude ramp, which leads into PWM (you can do (linear, constant frequency) PWM with a single 556, but I can do it with a single LM393).

By and large, I don't need constant frequency square waves; when it comes to switching supplies, most of mine look like either a two loop PWM system, or a variable astable with peak current mode control (something like a UC3842, with the oscillator supply strapped to the COMP pin -- so frequency is variable, keeping idle current down).

Tim

I love your website man! You've got some really nifty circuits on there. Once I get out of the digital domain, hopefully I'll be able to design a few beauties too.

If you've got a big budget then the Linear Tech TimerBlox are quite nice,

PIC10F series 6 pin Sot23 are pretty neat for this kind of thing. But then you have the added complexity of added firmware

I've quite regularly looked into the PIC10F series though I don't have an RS232 programmer or any PIC design stuff. I've got an STM32F050 TSSOP package in breadboardable mode too. The STM32F030 is only 70p a piece too which I think is just remarkable! Thank you however for the input