What are these voltages measured relative too? Where is the ground lead of your scope probe connected? If part of your ground net (with the probe tip) is reading -5v relative to another part (with the ground clip) you have a serious problem! Are there even any negative power supplies in the system?
If it were only the VCC dropping, it would seem like an inrush current issue. I'd try another smartmedia card if one were available. Also would considering putting together something to power up the card with no signals connected, with a small series resitor in the ground lead to it, and measure the voltage across that resitor, hence the startup current. You should expect some startup current, which would be supplied by capacitors located close to the card socket, but it shouldn't be huge.
But what you are seeing in the ground suggests something more systematic is wrong with the circuit or your measurement setup.
rebuild the power supply, or find out why the SM is drawing enough current to put the PSU into shutdown. What is the maximum current the supply can deliver? What are the startup parameters for the SM card Are the 3V3 componets 5V tolerant?
Hi All, I enconuter a big problem, when I insert a SmartMedia memory card to my system, my system power will be shifted, the VCC will be shit down to ground abruptly and then reutrn to VCC, the ground is doing the same thing, from 0V to -5V.
my system ( embedded system ) will crash after that, in my case, it will auto reset ( restart) when I insert SmartMedia card, but it will not happen every time, but it happens.
I main chip is running 5V, other components is running at 3.3V.
I missed this at the first look at your post You are measuring it _incorrectly_. As cs_posting suggested Check your grounding carefully. There should be no V on the ground connection. Or You have the scope probe ground on a point that IS NOT GROUND.
I didn't go to make sure the 3V3 componets have 5V tolerant or not yet. Due to I can see the supply voltage will become lower than 5V( maybe 2V,
3V..whatever), I think that is a big issue, because my main chip (CPU) is woking on 5V supply voltage, the others DRAM/UART/xxxx/... I think it maybe not so important at this time, how about DRAM? will DRAM cause system reset ?
current is a good point.
one possible thing, the system will not( or very few) be reset before I add EMI circuit/beds..
and I measure the ground relative to the embedded system's ground.
The crocodile(?) clip of picture is connecting to system's ground, and the probe is connect to the ground of components. ( they should the same right?@@)
I use one probe to measure the 3V3 and the other one to measure GND...
Now that the picture will display (at least sometimes) your problem becomes more clear. Your 2Gsample scope is fast enough to capture the full ugliness of the few nanoseconds long transient that results from plugging in the sim card, and the subsequent ringing as it bounces around for a total of perhaps 15 nanoseconds. A classic 20mhz bandwidth hobbyist scope would probably show none of this.
But the fix is simple. Modern electronics do nasty things like this every time they switch, not just on power on. CMOS is particularly bad
- it consumes almost no power sitting there, but when it switches it practically shorts out the power supply, but only for the briefest instant.
The solution is to provide a miniature power supply for each part of the circuit, located immediately adjacent to the device it services so that there are no resisteive or inductive imdepiments to providing the needed power immediately. This miniature power supply is called a bypass capacitor. In the days of through-hole components, you'd see a little .1 uF cap placed right at the end of each IC. They are still there in the surface mount era of course, but harder to identify. In the vicinity of the clock edges, those tiny little bypass caps are
*the* power supply for the circuit, while in between the clock edges they are recharged from the larger filter caps and the external power source of the circuit.
So here's the fix:
- Place a bunch of low series resistance .1uF caps right at the card socket, as physically close as possible.
- Place some larger, 1-10uF low series resistance caps somewhere close by.
- Ideally redesign the board with ground and power planes, or at least short fat power traces to minimize their resistance and inductance for feeding everything else on the board (though you may want some impedance in the supply to the card socket, see below)
One might think that the bypass caps belong in the removeable card itself. And there probably are some in there, but they are actually part of the problem you are experiencing: when first plugged in, those uncharged caps constitute dead shorts across the power supply! You have to have enough charge stored in caps right at the socket to fill them up, so that they don't send undervoltage pulses bouncing through power traces that would constitute a nightmare in terms of radio-frequency analysis.
If none of that works, a more extreme measure would be to power the card socket and it's bypass caps through a series resistor or inductor to limit that startup current. If this causes problems in operation of the card, the final extremity would be to soft-start the card through such a series limiting element, and then close a switching device to take it out of the circuit a few ms later.
Oh, and it's not just the card socket that needs bypass caps - every other IC does too, often a handfull in the case of a modern processor.
1=2E The shifted VCC is huge, but it seems that not the main reason about system reset. ( sometimes I got huge shift between VCC and GND (cross...), but the system still work stable..)
Question 1: The vabiration time is about 5ns~10ns, is that a safe time for .25um process or .5um process CMOS ?
2=2E We put all ground together... it seems that a big problem. I will cut the Card reader part tomorrow.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.