Reducing the Ring on a 555 Square Wave

Have you tried a small capacitor to ground?

If you are in sockets possibly try a small series resistor followed by a small capacitor to ground.

Just wild guesses.

Thanks Bruce. I was wondering if that would do it. I can calculate the RC time constant I need pretty easily for that. I'm giving you a GA for that!

Yes, yes, thank you. I need to drink some more coffee. I've used these tricks before and forgotten how sensible they are.

I was just thinking about how I could make a 3rd order or 4th order Butterworth filter out of the Op Amp downline from the 555. And yes, it is a chopper design. Old, but kind of cool because of how well it reconstructs the original waveform.

Thank goodness for this forum. My brain is too stuffed with trivia for keeping all of these good ideas inside.

I was more worried about the effect on the DC power bus from pumping the signal transformers too hard. I suppose the PWB is what I call a PCB and yes it is two sided.

I forgot to mention that my square wave was only about 6.6 KHz but the signal transformers give me lots of inductance as well. The older circuit I am modeling has a very clean square wave but it is 40+ years old. That vintage of chips is quite different from the high frequency devices made today.

At this low frequency, its all RLC reaction that I'm trying to tame. Thank you for your contribution.

I fought a similar problem with a microprocessor product. The 74138, 1 of 4 decoder used as a chip select, went from a part that was 8ns rise fall to something in the 100ps range. This caused the chip select line to go below ground over 5 volts, as well 5 volts above the Vcc +5 supply. This negative voltage was into a UART. Everytime the part was polled to see if a data word was available, it caused a parity error, which then slewed the current data word being received. It was some sort of internal IC problem that did not like current being pulled from the substrate (input pin intrinsic diode to the substrate). I found all the data and address line buffering chips to have this same problem, with them I could find a pin compatible part that had a 25 ohm output resistors on the totem pole output driver. I could not find the same for the 74138, but I change them to an older logic family, which I had to hard code into the BOM (bill of material) as well specify a part with notes to not substitute. This design had a 6 layer board, but the kid that designed it was new, as well this speed of processor (6Mhz) so no attempt at signal integrity existed in the board. I didn't have the luxury of redesigning the PWB, as this was certified avionics, and too much EMI/EMC as well temperature testing would be required for a very old design that was in support mode. My point here is not the clock speed, but the edge rate that caused all the issues. If you keep Lenz's law from clobbering you, ...... The intrinsic inductance of circuit traces is larger then you would think, especially when driven with picosecond pulses.

I should add that all these replacement parts met the original specs perfectly, the rise fall was faster then the minimum original component specification. I have no idea how fast a modern 555 timer is, but as they keep changing the silicon technology to build them along with fast parts, me thinks it's damn fast these days.

Faster chip drive circuitry is a common problem. Often adding a simple terminating resistor at the receiving end to make the transmission load more real than imaginary will do the trick without sacrificing rise times. One will lose some amplitude but isn't the high amplitude the root of the problem.

Want 1/4 million components? very cheap.

bazzer

I think I understand what you are trying from your description. (Schematics help so very much.) Decoupling the fast op-amps with supply capacitors might make it worse, It might make ringing less. It could also make no difference at all. If the op-amps are on the edge of stability then a host of possible problems may exist. I would just slow the op-amp down with a capacitor in the inverting feedback path.

Wait a minute. A 555 timer can easily produce a slow supply rail to supply rail signal. No gain will be needed then. A proper isolation transformer should attenuate an in band signal with out a problem. Square edge ringings will likely be out of band.

I take it back, I don't know what you are trying to do. Post a schematic, please.

I'll second ignator's suggestion to add a series resistor inline with the signal at the 555 output. As your signal is small, start with a small value, 22 ohms or so. That should give you some improvement.

Thanks to everyone for your suggestions. Here is what I did to resolve the issue.

After burning up a few 555's due to a duty cycle miss-adjustment I found that I could reduce part of the ringing with a ceramic bypass capacitor in parallel with a large nearby electrolytic capacitor. (A classic bypass trick, I think.)

To get the single ended signal to drive a push-pull amplifier, I drove the output through a 10 uF ceramic capacitor and then amplified it with an offset pot to create the required double ended (±12V) drive for the system. I also placed a small ceramic capacitor in parallel with the feedback loop which put a little slope in my rise time, but it wasn't enough to cause a problem.

With this being a surface mount project, space was already tight. I have cooked too many 555 timers to experiment with additional load, so capacitors were my first choice. All I had to do was figure out where to put them. I'll have to redraw the circuit to post it because of the likelyhood of lost resolution. Next post, I promise.

Here is the circuit:

Shouldn't the PULL transistor collector supply be connected to GND? I'm confused how this works. Seems the LED4 diode is pointing at this problem as well.

What is not shown is the coil (a primary) of a signal transformer to ground. Each transistor takes turns turning on with the ±12 V square wave. That way the coil sees a peak to peak voltage of 24 volts. But the issue seems to be getting enough current through it without melting the 30 gauge wire of the transformer.

Later in the circuit, the imposed square wave becomes the power supply for some JFETs that are driven by the signal to be isolated. Finally, the signal comes out reassembled and is then amplified. When I first looked at it, I had a hard time figuring out how the JFET transistors did anything, but it really works. The original signal is less than 0.15 Volts zero to peak. Quite a challenge!

Alas, once again the image decimation of CR4 has rendered a schematic unreadable to me.

This circuit looks far too complicated for my preferences. Then again I still don't grasp what you are doing with this circuit outside of hard driving an apparently very lossy signal transformer. I'm glad it works for you.

The original bipolar 555s take a very fast spike from supply when both + & - output transistors conduct during the switch-over - very good bypass close to chip was recommended to make supply "hard".

That offset amp may be much faster than original, passing ringing a 741 would not.

There is a time in the middle of the drive swing when neither pull or push is on. At this time, the inductive current in the transfo will find another path. It may have pulsed in the secondary of transfo.

This kind of push-pull output is used in audio amps, driven by a single transistor with resistive collector load, a circuit of resistors & diodes between the bases of push & pull sets enough standing current to avoid "cross-over" distortion.

Synchronous full-wave system with transfo isolation is very effective to avoid noise, I remember them in intrinsically safe analog isolators for gas explosion hazard.

That offset amp positions the output at middle OK but with low output Z of 555 & 2M2 feed back, it probably has far too much drive & gain, which is a way to make it ring.

You have to be very clear to the 555 that you are just friends and are not looking for a long term relationship. Then the ring shouldn't be an issue.

Hahaha. You funny!

The original circuit may have never worked correctly (by design), since all the old-style 555 timer datasheets I can find rate the part at an absolute maximum operating voltage of 18V. your circuit powers the parts with 24V. Any parts that don't burn up immediately may not have very long lifetimes in this circuit.

If you're looking at the circuit in post #14: then the 555 itself is powered from +12V and GND; only the subsequent circuit uses the -12V rail.

Oops! My mistake. I should have looked at the fuzzy graphics on a larger screen. Please ignore my post.

Thanks,
Carl

Oops! My mistake, the 555 is only running on 12V.

A .1 uF ceramic on power rail as close to IC as possible will most likely fix it. But different manufacturers' 555s do behave differently on fast applications so trying other brands won't hurt. Also generally CMOS 555s like LMC555, ICM7555 or TS555 are much faster than bipolar ones and don't shake the power rails too much. But note that max operating voltage and available source and sink currents are smaller so all depends on your (unreadable) app. S.M.