ringing in logic chip

[UtrasoundGuy]

this should be a simple question as its only a 1 (maybe 2 part) logic device.

in this circuit, U54 is an AND gate and which combines a DC signal (as far as we are concerned it wont switch) with a negative pulse.
The negative pulse goes from 5V to 0V for 500ns and then goes back to 5V for the remainder (around 2.5KHz pulsing).
Sometimes the pulse is alittle below 0V (due to a ground loop) but that shouldn't be an issue.

http://files.engineering.com/getfil...695-4b6e-b280-843d0c159c32&file=schematic.bmp

we have the following inputs and outputs of U54.
pink is u54 output (TGC trig), green is scope on U54 pin2 (input), blue is a few feet back on the cable to pin 2 (input)
we are not scoping the pin 1 input to the AND gate which is a clean 5V (ripple less than 0.1V)

http://files.engineering.com/getfil...a646-4282-88b6-c74fcbb1f78a&file=scope_57.bmp

So why is the pink output ringing so much.
Ive tried to filter the input with a 200ohm resistor in series (with respect to input and U54 pin 2) and 200pf cap from u54 pin 2 to ground.
This has little effect on this noise/ringing coming out of the chip.
Ive also disconnected all the outputs of the chip that go to other PCBs so that no external effects can come in.

I can put a cap on the output and clean it a little but I think there is something wrong that is the root cause.
This is a 15 year old design in a console a electronic module with 1500 parts but the problem is with this one chip.
Im guess impedance matching could be an issue?

thanks

 

[UtrasoundGuy]

I forgot to say, if we scope U53 which is utilizes the output of U54, the chip output appears to be functioning correctly.
So it unlikely the next chip is broken and is causing loading.

This is not for one board, it is a design issue effecting all the boards.
for some reason, it has a history of problems with quality going back 10 years that no one has addressed.
The problems include U54 breaking.

thanks

 

[IRstuff]

You probably should have continued on in your original thread. Again, You keep saying that it's broken, but I can't see where you fully explained how it's broken, which is crucial to figuring out the problem. You've removed a number of parts, so has anyone done failure analysis, or even attempted to see what's actually broken? Is it the input or the output? Did U53 work on the boards that were "broken?"

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[Skogsgurra]

I noticed that other thread about gate failures, but I didn't read it in full length. So it is possible that this has been mentioned before.


Anyhow, the ringing is violent and the frequency is somewhere in the 10 - 20 MHz region. Say 15 MHz. It does not look like an impedance mismatch (reflected wave) problem.

The amplitude is under 2 V (scope input saturates) and the substrate diodes should have shorted it out long before that.

There are 10 uH chokes and .1 uF and 1 nF decoupling capacitors in the Vcc supply.

Thoughts: Some capacitors and chokes do ring. Have you looked at the Vcc? Does it ring?

The amplitudes are unrealistically high. Question; is the pink probe properly compensated? If not, you could easily have your ringing "amplified" by the probe.

Then, do you have the same ringing with an older generation of gates? Gate geometries are shrinking and that means that it takes less energy to destroy them. The heat generated by the ringing may be sufficient.

Sorry if these thoughts already have been presented. What, in that case, were the answers?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[UtrasoundGuy]

In some ways, this is a new issue. The last issue had to do with the input to the chips and why they were breaking. This is the output of the chip and it seems to occur independently of what is done on the input side (filtering it or not).

IRstuff - for practical purposes, that is the entire circuit. The signal comes into U54 from a galil controller board (CMP output). It leaves the AND gate and goes to U53 but we are not using the output of U53, but i did check it is working correct. The signal leaves U54 and also goes toa differnt module, a 74LS123 (monostable vibrator). If disconnect it from that 74LS123, i still get the same level of ringing.

The entire schematic for this one board is 22 pages long and for practical purposes, and there are 6 boards like this.

Last night i bread boarded just the one chip (74ACT08), attaching short pieces of wire wrap to the surface mount leads.. I used a function generator to simulate the galil controller pulse (5v to 0v for 600ns) and a voltage regulator (5VDC) for the other input. I am getting the same response.

I recently just bread boarded the one chip (74ACT08) using a fucntion generator as the input (5v to 0v at 600ns) and 5V dc as the other input. I get teh same ringing on the output with this one chip alone. I can put 300ohm in series with the output and it reduces the ringing to a normal pulse. Im not sure why as when i test it, the 300 ohm is not going anywhere... one end is at the output of the 74AC and the other just goes to the oscilloscope which in theory has a meg or 10meg resistor in series with it. So why is it reducing the ringing?

Skogsgurra - already presented that stuff. The VCC is clean and has chokes on it. The ringing is from switching.

 

[UtrasoundGuy]

with 300ohm in series to oscilloscope probe (green is output, yellow input to FG)

http://files.engineering.com/getfil...fa2c-4430-ab9c-4b9eb040d994&file=scope_59.bmp

without 300 ohms in series to oscilloscope probe (green is output, yellow input to FG)

http://files.engineering.com/getfil...a2fc-4219-9879-15656cd63387&file=scope_60.bmp

 

[Skogsgurra]

You didn't get it. But let's forget about the cause of the ringing for now.

Also, there's more to the probe than the 10 Megohm. I asked if the probe was properly compensated. Is it?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[UtrasoundGuy]

I frequency do the probe calibration that is built into the Agilent oscilloscope we are using.
I had did one yesterday when I saw this output due to confusion.
Many the calibration is not at the right frequency, its a built into the scope thing..

I haven't tried with an old generation of AND gate yet.
Ill see if I can find one.

 

[IRstuff]

Did the board EVER work? You sort of imply that the 74ACT08 is a new thing in your last post. The ACT generation is characterized by faster transition rates, and dropping it willy-nilly into an older board layout could result in ground bouncing, particularly on outputs where both pullup and pulldown gates are briefly on at the same time.

For that matter, is it a real functional problem? You say it goes to a one-shot, and while the ringing is obnoxious, the worst it can do is re-trigger the 74LS123 which might stretch out its output. Have you looked at the output of the 74LS123? The fact that U53 output shows no problem suggests that there is no problem to investigate.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[Skogsgurra]

It is also important where you put your ground clip. Connect it to the gate's ground pin.

I cannot imagine that the ringing you see shouldn't be clamped at between -.7 and -1V. Substrate diodes always clamp just below zero. That is why I keep asking about probe compensation and also where your ground clip is connected.

One question; your nick says "Bioengineer" - does that mean that you didn't have any electronics training? In that case, I think you should get hold of a seasoned electronics engineer. He will sort this out for you.

BTW, the built-in probe compensation doesn't need to be of any particular frequency. The main thing is that you get your corners "straight".

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[UtrasoundGuy]

skogsgurra - I appreciate that you are trying to help me out.. However, that last post was very insulting. Your comments are useless to me as they are very basic nor you dont seem to take the time to read the facts that presented, all of them (such as the ground loop and unplugging the NI card fixes the issue, or using differnet ground points or even low inductance copper ground plane to enhance it). Please dont respond to anymore of my questions, ever.

IRstuff - The 74ACT08 and 74LS123 worked in the design for 10 years. Its not my design. Ill check the output of the 74LS123 a little later.. Since I have to fix the problem with them breaking from the input, Im want to correct the output ringing too. Im debating if it needs to be done. The fact is that if i measure on different sides of the reistor, i get different peak to peak voltages. I can get teh same effect with a resistor or cap to ground. From what I see, Im giong to have to add a resistor, but want to know why first.

 

[Skogsgurra]

I am very sorry if you are insulted if I try to be helpful.
I cannot help thinking that you need to have someone that knows about electronics to help you. The negative excursions are one thing that most probably is a result of either the ground clip being located badly or the probe not being compensated. Your remark that "the calibration is not at the right frequency" indicates that you may be on deeper water than you realize.
Sorry again.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[IRstuff]

"such as the ground loop and unplugging the NI card fixes the issue"

This is the first time you've mentioned this. "unplugging NI card fixes the issue" says the AND gate output is a red herring and merely a symptom of a more global problem.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[IRstuff]

A dirty fix would be to put an optoisolator in the path, which ensures that the signal grounding is decoupled from this signal.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[UtrasoundGuy]

sorry if I didn't mention it earlier, but this console that seems have a ground loop. I thought I had said this. The DC steady state region of trigger pulse going into the AND gate goes to as low as -400mv. This pulse is also effected by negative overshoot as much as 1V as was mentioned in the previous post and as that overshoot dies down, you get to the DC state.. For the DC component, when I disconnect various parts of the console, I can get a connector from NI that causes the trigger voltage (to AND gate) to go back from 5V to 0V rather than 4.6V to -350mv. The NI card has nothing to do with motor control, just PCA functionality. Its going to be a beast to fix but I haven't even started yet. Its more of a symptom. But for now, I have been concentrating on the overshoot which really has been fixed, im looking at the output now which is a different topic, somewhat.

For now, Im going with the resistor between the two chips. Its a much lower value though than what we prototyped, just 100ohms needed vs 300 on the bread board. This ringing is definitely effecting the output of the next chip. Although there is no DHR, im guessing that is part of the reason they put a monstable vibrator in as the next chip as that there is no ripple that comes out of it when the ringing goes in... Although this next chip is not breaking, we are using it out of spec in that the pulse is going to -2V. 100ohms seems to be all it took but I have to do a full console verification tomorrow.
.

 

[UtrasoundGuy]

id still like to know why the resistor on the dead end of a IC (only going to the probe) would effect its output measured on both sides.
There should be no current, I guess it might just dampen the alternating current sharp pulses
I dunno.
thanks all

 

[Skogsgurra]

Of course there is a current! The probe's input capacitance has a lot to say when the edges are at about 10 ns. Input capacitance can be around 3 pF.

Five volts in 10 ns results in 1.5 mA. So, your observation that a resistor at a few hundred ohms damps the ringing is not at all improbable.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.