Hi On some of the 21161 based boards in my application I see a problem such that the chip does not power-up properly every time. I don't know what exactly happens, but the DSP is not running as expected and it *can't* be brought out of that state by a hardware reset. Power cycling is the only way to cure it. Does anybody have an idea what that could be? I checked all the unused pins for proper "treatment". Regards, Daniel
21161 power-up problem
Started by ●September 13, 2005
Reply by ●September 13, 20052005-09-13
weiss@weiss.ch wrote in news:1126644968.880896.205990 @g49g2000cwa.googlegroups.com:> Hi > > On some of the 21161 based boards in my application I see a problem > such that the chip does not power-up properly every time. I don't know > what exactly happens, but the DSP is not running as expected and it > *can't* be brought out of that state by a hardware reset. Power cycling > is the only way to cure it. > > Does anybody have an idea what that could be? I checked all the unused > pins for proper "treatment". > > Regards, > Daniel > >Do you know the rev of the chip? -- Al Clark Danville Signal Processing, Inc. -------------------------------------------------------------------- Purveyors of Fine DSP Hardware and other Cool Stuff Available at http://www.danvillesignal.com
Reply by ●September 14, 20052005-09-14
Chip revision is 1.2 (I guess that is the latest?) Daniel Al Clark schrieb:> weiss@weiss.ch wrote in news:1126644968.880896.205990 > @g49g2000cwa.googlegroups.com: > > > Hi > > > > On some of the 21161 based boards in my application I see a problem > > such that the chip does not power-up properly every time. I don't know > > what exactly happens, but the DSP is not running as expected and it > > *can't* be brought out of that state by a hardware reset. Power cycling > > is the only way to cure it. > > > > Does anybody have an idea what that could be? I checked all the unused > > pins for proper "treatment". > > > > Regards, > > Daniel > > > > > > Do you know the rev of the chip? > > > > -- > Al Clark > Danville Signal Processing, Inc. > -------------------------------------------------------------------- > Purveyors of Fine DSP Hardware and other Cool Stuff > Available at http://www.danvillesignal.com
Reply by ●September 14, 20052005-09-14
weiss@weiss.ch wrote in news:1126679496.747798.192440@o13g2000cwo.googlegroups.com:> Chip revision is 1.2 (I guess that is the latest?) > Daniel >The early 21161 had a problem with its PLL. This is corrected in the revision that you have which I believe is current. I assume you have some kind of reset part connected to reset (not just an RC circuit). Here are ideas: 1. You may have a rise time problem with your power supply. 2. You might also look at the decoupling. 3. You should also have a Schottky diode between the core and 3.3v supplies. These kind of problems are usually not the chip but something else. We have never seen this problem with our 21161 designs. -- Al Clark Danville Signal Processing, Inc. -------------------------------------------------------------------- Purveyors of Fine DSP Hardware and other Cool Stuff Available at http://www.danvillesignal.com> > > Al Clark schrieb: > >> weiss@weiss.ch wrote in news:1126644968.880896.205990 >> @g49g2000cwa.googlegroups.com: >> >> > Hi >> > >> > On some of the 21161 based boards in my application I see a problem >> > such that the chip does not power-up properly every time. I don't >> > know what exactly happens, but the DSP is not running as expected >> > and it *can't* be brought out of that state by a hardware reset. >> > Power cycling is the only way to cure it. >> > >> > Does anybody have an idea what that could be? I checked all the >> > unused pins for proper "treatment". >> > >> > Regards, >> > Daniel >> > >> > >> >> Do you know the rev of the chip? >> >> >> >> -- >> Al Clark >> Danville Signal Processing, Inc. >> -------------------------------------------------------------------- >> Purveyors of Fine DSP Hardware and other Cool Stuff >> Available at http://www.danvillesignal.com > >
Reply by ●September 14, 20052005-09-14
Yes I have a reset circuit connected. The reset timing is fine. I also thought of the power risetime, but that seems also fine. The Schottky diode is also there. I'll try some more decoupling. I have to say that 4 boards out of 50 show that problem, so it seems to be some design parameter being at the edge. I also doubt that it is the DSP chip, because 8% bad chips would be a lot. I use a crystal for the clocking. The starting period of the oscillator seems also fine, well within the reset active state. Thanks for your help - if you can think of anything else let me know. Daniel
Reply by ●September 19, 20052005-09-19
I have found the reason in the meantime. It is the sequencing of the 1.8 and 3.3 V supplies. In my application both voltages slew at the same time. If I delay the 1.8V for about 2ms, all is fine. The Schottky diode you mention takes care that the 3.3V do not lag behind the 1.8V, but it seems that the 21161 prefers to have the 1.8V lagging a bit. The whole is temperature dependent, i.e. when the DSP is cold it requires a higher lag time to power up reliably. Also most of the chips I had had no problem in that respect, it is only with a few. Daniel
Reply by ●September 19, 20052005-09-19
weiss@weiss.ch wrote:> I have found the reason in the meantime. It is the sequencing of the > 1.8 and 3.3 V supplies. In my application both voltages slew at the > same time. If I delay the 1.8V for about 2ms, all is fine. The Schottky > diode you mention takes care that the 3.3V do not lag behind the 1.8V, > but it seems that the 21161 prefers to have the 1.8V lagging a bit. The > whole is temperature dependent, i.e. when the DSP is cold it requires a > higher lag time to power up reliably. Also most of the chips I had had > no problem in that respect, it is only with a few.Thanks for the heads-up. That kind of inside information is very valuable. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●September 19, 20052005-09-19
weiss@weiss.ch wrote in news:1127162145.573764.24630 @g44g2000cwa.googlegroups.com:> I have found the reason in the meantime. It is the sequencing of the > 1.8 and 3.3 V supplies. In my application both voltages slew at the > same time. If I delay the 1.8V for about 2ms, all is fine. The Schottky > diode you mention takes care that the 3.3V do not lag behind the 1.8V, > but it seems that the 21161 prefers to have the 1.8V lagging a bit. The > whole is temperature dependent, i.e. when the DSP is cold it requires a > higher lag time to power up reliably. Also most of the chips I had had > no problem in that respect, it is only with a few. > Daniel >I've never seen the problem you mention, however in our design, we created the 3.3V and a 5V supplies from a higher DC voltage. The 1.8V core was then derived from one of these supplies. This means that our 3.3V supply was always on first. On the 2126x, we use a dual switcher to create 3.3V & 1.2V. We can assume that both supplies are valid at about the same time. I have never observed this problem with these parts, but then again, we are talking about a significantly different design. It doesn't surprise me that temperature is a factor. I have seen some very strange temperature anomolies. The craziest was a crystal that would only fail during a temperature gradient as it went thru about 120 degrees F. It worked fine at both lower and higher temperatures. A significant number of crystals failed in this manner. It wasn't lot dependendant either. We changed from a HC-49 short to a HC-49 tall xtal from the same manufacturer and never saw the problem again. (BTW, this was before Danville) -- Al Clark Danville Signal Processing, Inc. -------------------------------------------------------------------- Purveyors of Fine DSP Hardware and other Cool Stuff Available at http://www.danvillesignal.com