Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

74AC Logic Ground Bounce & power

Status
Not open for further replies.

dr pepper

Well-Known Member
Most Helpful Member
I havent used AC logic a lot.
However I need to divide 125Mc down so I have a couple of these chips.
I prototyped the circuit on breadboard with 100n's accross each chip and it worked.
Just built the circuit onto a vero type board also with 100n's across the chips and it doesnt work, at all.
So I held a 10u cap across one of the flip flops and then it works ok.
There is about 0,5v sinewave at the clock freq across the chips power lines without the 10u, not much with it.
The chips power goes to a bus made from 1mm copper wire with 0.2mm wire wrap (soldered) about 1"away.
Thinking just solder a electro accross each flip flop chip.
 
74AC it totally evil as far as ground bounce goes. The switching edges are extremely fast, and the power lines are in the corners giving the largest possible inductance. I have seen over a volt of ground bounce on a non-switching output.

That was why the 74AC11000 series was invented, with twin power leads (usually), and power leads as close to the chip itself as possible. However that series never really caught on and there isn't a good range available.

I would avoid corner-pinned DIL ICs at 125 MHz. Alternatives are a https://www.ti.com/lit/ds/symlink/74ac11074.pdf or a https://www.ti.com/lit/ds/symlink/sn74lvc1g80.pdf

The 74LVC1G80 can be wired to divide by 2 by wiring Qbar to D. The input goes to CLK, output from D Qbar, and connect the power. Maximum frequency is 160 MHz.

Farnell have them in stock and I think I have a few hundred somewhere.

If you want to divide by 10, divide by 2 first, and then by 5. Dividing by 5 is a lot easier at 62.5 MHz than 125 MHz.
 
OK hadnt heard of those.

Not having read your post and messing further I got the thing to work, it seems the problem was the sockets, the chips are in cheap pressed sockets, thats obviously a no no, so as a temporary measure which might well end up permanent I soldered the power pins in to the socket, I get a reasonably nice o/p waveform,, if I had some 'HC ones I'd put those in the latter stages.

I'll look at the LVC range next time I try something wacky like this.

Your right about the edges and harmonics, I have a gps receiver that gets wiped out by it, 5 feet away, I had to put ferrites on the cable to get it to work, but it shouldnt be bad if I build the divider into a old IF can.
 
Hmmm... My experience with AC logic is just the opposite. I wish everything were available in AC. Maybe it likes american-style football better, and is acting out...

At 100 MHz, sockets are a no-no no matter what the logic family is. The extra conducting path length is only a couple of mm, but that's enough to cause problems. Also, at that freq a ground plane is pretty much mandatory for repeatable performance. Everything is about loop lengths. For example, when positioning the 100 nF decoupling cap near pin 14 or pin 16, orient the cap parallel to the top of the chip, with the Vdd pin near the chip pin and the ground pin toward the center of the chip. This shortens the effective loop length through the ground plane to the GND pin. At the frequencies necessary for reasonably square edges at 100 MHz, the diagonal distance from pin 16 to pin 8 is over 10% of one wavelength. That's enough to disturb most rules of thumb about digital logic layout.

ak
 
AS series are 50% faster than AC- perhaps beneficial at that freq...

i wire .o1uF mylar films across all my hi speed logic chips to keep them happy...
this is assuming you have smooth supply
 
Last edited:
AS I'll have to look them up, never used them before.
I think I'll also use turned pin sockets on anything fast, these were cheapo's:

IMG_20160413_213000.jpg
 
I'll be surprized if changing the type of socket construction has any real effect. Usually, ground-bounce problems come down to inductance, and while one socket construction's inductance obviously will be slightly different from a different construction's inductance, the minor variation in value will be swamped by the overall value. As a quickie way to test this, tack a small ceramic capacitor across the device power and GND pins *on the chip* with the shortest possible leads. In this way, the decoupling cap is "inside" the lead and socket inductances.

ak
 
Usually, ground-bounce problems come down to inductance
Often the ground bounce seen is from the inductance of the scope ground lead as well. Using a low inductance probe is important in a high speed signal, where signal quality is being tested.
When working with high speed logic such as here, one should use RF layout techniques. The sockets are all sort's of parasitic, so predicting the final circuit behavior would be a challenge.
 
Yes.
I have a 'better' 'scope probe, it doesnt look much better with that.
The circuit now works well with the power pins soldered in, I probably should have put the ic's direct in the board.
To be honest I'm amazed it works at all, the osc is a 125mhz sine wave, I just biased the 74ac74 with 2 resistors to 1/2 supply and stuffed the i/p to that through a 100n cap.
It might have worked better if I'd used some turned pin sockets with built in power supply bypass caps.
 
Yeah, the sockets with integrated decoupling are great for this kind of stuff. Whenever I needed them I always had more time than money, and made my own.

ak
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top