SX48BD (Parallax label) works at 5.0Volts, not 3.3Volts.
dkemppai
Posts: 315
Ok,
This one is bugging me. I have a chip, and program that I've been working on for some time now,
and have all of my software is currently working. However, when I try to power the chip at 3.3Volts,
I get some really strange behavior.
First off, BOR = 2.6 Volts. My 3.3Volts is from a good LDO, properly bypassed. I've checked the voltages,
and they are rock solid (I scoped them for anything, not even a glitch, nothing, nada. 500nS/Div Nothing.
This is not a voltage problem. Also, This everythign has good connections. This is not·a proto board, I had
a production run of tested multi layer boards built. No bad connections. Again, EVERYTHING works just fine
at 5.0 Volts.
My first problem at 3.3Volts was·resonator didn't want to start. I tried a bunch of things, including a call
to tech support. Tech support suggested a 5pF Cap to ground from OSC2 (This is new to me, you guys ever
heard of resonator problems needing a cap to ground?). That didn't do it. However, a 5pF cap
in parallel with a 1 Meg resistor to ground seemed to do the trick. At least the oscillator is running.
Now, the real problem. With the oscillator running,·my code will not work at·3.3Volts. It runs properly
at 5.0Volts, just not 3.3Volts.·Here's what I notice:
I have a pin that is serial data out. Whe I run the SX at 5.0 volts, the pin gets set to·hi (no data) shortly
after the oscillator starts up. When the SX us run at 3.3Volts, the pin never goes high. I set a few debug
pins, and attached them to the scope, and it appears that none of the outputs get set. It seems like the
SX is not·setting·mode registers properly (Just a guess).
Out of frustration, I changed the chip and the oscillator (Which is a (expletive) to do with these surface mount
components). Same behavior!
So, I'm out of ideas? What am I missing?
AGAIN, Please note, IT works at 5Volts, BUT NOT at 3.3, and Brown Out is set to 2.6 Volts
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
This one is bugging me. I have a chip, and program that I've been working on for some time now,
and have all of my software is currently working. However, when I try to power the chip at 3.3Volts,
I get some really strange behavior.
First off, BOR = 2.6 Volts. My 3.3Volts is from a good LDO, properly bypassed. I've checked the voltages,
and they are rock solid (I scoped them for anything, not even a glitch, nothing, nada. 500nS/Div Nothing.
This is not a voltage problem. Also, This everythign has good connections. This is not·a proto board, I had
a production run of tested multi layer boards built. No bad connections. Again, EVERYTHING works just fine
at 5.0 Volts.
My first problem at 3.3Volts was·resonator didn't want to start. I tried a bunch of things, including a call
to tech support. Tech support suggested a 5pF Cap to ground from OSC2 (This is new to me, you guys ever
heard of resonator problems needing a cap to ground?). That didn't do it. However, a 5pF cap
in parallel with a 1 Meg resistor to ground seemed to do the trick. At least the oscillator is running.
Now, the real problem. With the oscillator running,·my code will not work at·3.3Volts. It runs properly
at 5.0Volts, just not 3.3Volts.·Here's what I notice:
I have a pin that is serial data out. Whe I run the SX at 5.0 volts, the pin gets set to·hi (no data) shortly
after the oscillator starts up. When the SX us run at 3.3Volts, the pin never goes high. I set a few debug
pins, and attached them to the scope, and it appears that none of the outputs get set. It seems like the
SX is not·setting·mode registers properly (Just a guess).
Out of frustration, I changed the chip and the oscillator (Which is a (expletive) to do with these surface mount
components). Same behavior!
So, I'm out of ideas? What am I missing?
AGAIN, Please note, IT works at 5Volts, BUT NOT at 3.3, and Brown Out is set to 2.6 Volts
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
Comments
As an added note,·this SX48BD WILL NOT operate below 4.89 volts.
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
-Phil
running. I'm monitoring it with a scope. 49.98Mhz.
Resonator is 50Mhz. Surface mount Murata purchased from Parallax.
10K feedback, 5pF to ground from OSC2, 1Meg from OSC2 to ground.
I think something is wrong with the·I/O port. It·appears be broke when
operating·at 3.3 Volts.·Again, when the chip starts up from 5 volts, my
serial data output goes high. When the chip starts up at 3.3 volts, that
same serial data output pin stays at ground. 3 chips in a row. Nothing
hooked to the pin when I test this (Only a scope).
-Dan
FPGA's are easier than this freaking chip.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
First you need to know if the chip is running or not, just because your serial pin stays at ground doesn't mean it stopped working.
Just make a simple LED blinker with the same DEVICE parameters and see if it works. If it doesn't post the LED blinker program.
You may need a different OSCxxx setting for 3.3V on the device line.
Bean.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I know what I know, don't confuse me with the facts...
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
www.hittconsulting.com
·
Done that. But I didn't use an LED, I used a 4 channel Tektronics scope.
I'm scoping the oscillator pin. I can see the 50Mhz. So, You're telling me that
just because I can see the 50Mhz, the chip may not be running?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
Just make an LED blink once per second, do see if the device is running with the settings you are using.
DO
HIGH RA.0
PAUSE 500
LOW RA.0
PAUSE 500
LOOP
Bean.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I know what I know, don't confuse me with the facts...
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
www.hittconsulting.com
·
I think what Bean was saying about the resonator is that the OSC setting you've been using when running at 5v may need to be changed when running the chip at a lower voltage like 3.3v. That should be a relatively easy thing to try and wouldn't take long to try all the different OSC combinations to see what one works best.
In regards to the way the port is working (or not), how have you set up the port? Have you tried the different combinations (pull-ups, CMOS, TTL, Schmitt, etc) to see if any of those are a factor?
Just for testing purposes, can you set the brownout even lower or disable it?
How are you handling the RTCC pin? Is it tied high/low or left floating?
How is your RESET line connected?
I've had really good luck using the SX chips for several projects and i'm sure you'll get it working.
Best Regards,
Robert
I've tried Several chips, and all·have trouble at different voltages.
ranging from 3.54 volts to 4.79 volts. The oscillator is always
running, in every case. I'm checking that with a scope.
READ CAREFULLY: Now,·I run the voltage just below the point where
the chip won't set·my test output pin.·I verified this several times.
But, if I·hit the chip with freeze spray, it will start and set the
output pin every time.·Once it warms up to room temperature again,·the
chip won't start up. If I freeze the chip again then it will start again
until it warms up.·
I've repeated this test several times. It's the SX. Not the software.
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
Again, a scope probe on the oscillator pin proves nothing, because you are changing the circuit by connecting the probe.
It sounds like a resonator drive problem, if it is one of the OSC settings should work.
Bean.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I know what I know, don't confuse me with the facts...
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
www.hittconsulting.com
·
-Phil
(Note, I didn't attach the scope probe until
after I verified my code wasn't responding).
············· OSC pin at 3.3V······ OSC Pin at 5.0V.······· Output "Blinky"·at 3.3V······· Output "Blinky" at 5.0V
OSCHS3 -·Running················· Running····················Nothing···························· Blinking
OSCHS2 -·Running················· Running····················Nothing···························· Blinking
OSCHS1 -·Nothing················· Running····················Nothing···························· Blinking
OSCXT2 - Nothing················· Nothing····················Nothing······················ ······Nothing
OSCXT1 - Not tested
OSCLP2· - Not tested
OSCLP1· - Not tested
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
Of course, it'll run at·5, and not at 3.3 with bor = 4.2
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
How long are the leads/tracks from the SX to the resonator ?
Do you have one of the thru-hole 50 MHz resonators you can try ?
When you do have the scope probe on the osc pin, does the waveform change when you spray the SX with cold spray ?
This is a real stumper, usually changing the OSC setting or using the brown-out setting fixes most startup problems.
Bean.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I know what I know, don't confuse me with the facts...
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
www.hittconsulting.com
·
I have Very short leads (Less than 1/4 inch for each lead) .
The boards are SMT boards, with only room for the SMT resonator.
The waveform is the same when the chip is hit with cold spray or not.
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
Does your Tektronics blink program run at 3.3v with a lower frequency resonator?
My clue here is that most other processors (e.g. PIC) will NOT run at the same speed when you reduce the voltage and the one known application that has been mentioned here which DOES run at 3.3 is doing so at 8Mhz. The old SX chips from Scenix WERE rated to run at 50Mhz down to 3.0v, but those were also rated to run at 75 or 100Mhz at 5v. I would not be sure that the current Parallax produced chips are quite so magical.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
---
James Newton, Host of SXList.com
james at sxlist,com 1-619-652-0593 fax:1-208-279-8767
SX FAQ / Code / Tutorials / Documentation:
http://www.sxlist.com Pick faster!
Just a thought...... please remember that the oscillator continues to RUN even while the chip is in RESET, so as Bean states, a running oscillator is not proof of a running program.
Peter (pjv)
-Phil
Just a guess, Chris
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
I'm not scared of your robot. I'm covered by Old Glory (youtube)
The only thing "magical" is how the Scenix sales and marketing people rated any SX chips to run at 100 Mhz. At the time they tested the higher-yield portion of the die with a faster operating speed. That practice ended quite early as well. These chips were from the same die that were rated for 50-75 Mhz (don't remember exactly).
The only changes to the·chip don't relate to the die at all, but to testing and packaging. We've tested to a·more stringent specification,·marking is now Parallax instead of Ubicom, and RoHS-compliant versions are available.
I'll see if we can duplicate the problem Dan is experiencing.
Sincerely,
Ken Gracey
Parallax, Inc.
·
I just·switched my board over to 5V for the MCU, and left
everything else at 3.3Volts. (Which is a pain since I allready had
boards done, and I hate modding boards for a deliverable)
Ken, I know the chip hasn't changed in a long time. I tried my
very last scenix labeled part, and it was also prolematic.·Note,
however that it did operate·at a·lower voltage
than the others did. ROHS packaging difference perhaps???
Phil, I can't post the schematic. But, I'm sure there aren't any
layout issues.·I do this often enough to know how to avoid
those. [noparse]:)[/noparse]
Chris, I may try that if I have time, however the SMT 20Mhz
resonators are larger than the 50's by a bunch. I may try to
shoehorn one in there if I have time.
Peter, The chip isn't in a reset state. Other·functions in code
are operating properly when the particular output isn't funcioning.
The OSC is running, and many of the other pins are acting the
way they should.
James, I've been having the same·thoughts you were. I think
you're onto something about a voltage/frequency limit with
the chip.·The data sheet appears to indicate they will run at 3.3Volts
at 50Mhz, but that was subject to more testing at the time the
data sheet was generated. (Or at least that's
what I got out of it)
Here's why I think you're right:
Where I'm·having trouble is on an Interrupt driven dual RS232 VP.
I·have the output routine drive two separate pins, Identical data
is sent out each port. (One is a communications channel, the other
is a debug port. Each pin is set·and cleared at the same time.
One lags the other by only a few clock cycles. One pin is on·portD
and the other on PortC. When I lower the power supply voltage, the
port D pin stops responding, and the PortC pin acts normally!!!
Again, the PortD pin is problematic, when PortC is not. This only
happens when I have·A LOT of processor load and the voltage
is low. Cooling the chip causes the port D pin to respond accordingly.
Seems like some sort of·internal propigation/latching issue to me...
I will post more if I get a chance to do more testing. I·want to
try the 20Mhz resonator, and see if things operate·there at 3.3Volts
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
-Phil
It's a double sided·board. Each VDD and VSS pin are tied to the appropriate power supply leads, with a .01uF SMT603 capacitor bypassing each set of power supply leads·placed no further away than than 1/16th of an inch. The Resonator is under 1/4 inch away from the chip with with the feedback resistor 1/16th of an inch away from the resonator. In the cases where the extra components were added, they were added to the top of the resonator (for lack of any·other place to put them).
The reset pin has a 10K pullup resistor as required, and is place only 1/16th of an inch from the QFP.
The chips is sourrounded by ground plane on the top and bottom layers, with plenty of ground tie vias to keep the ground
plane impedance low.
The output pin in question resides on a circuit trace less than·3/4 inches long...
-Dan
P.S. I used the same general rules to lay out this board that I did with·+50Mhz LVDS signals and·faster signals·sorrounding
a 150Mhz·FPGA board with onboard DC-DC converter that I did earlier this year.
I'm pretty sure it isn't a layout issue,·Although I can't prove it [noparse]:([/noparse]
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
-Phil
This one is really bugging me!!!
If I can be assured of confidentiality, I may send a board and code into Parallax Tech support for testing, so they can verify my results. However, that won't be until after I ship a few units to satisfy my current contract.
Thanks all!
-Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
This just doesn't make any sense at all. These chips work, and that's all there is to it! That said, you still have a problem, so how can we help you get to the bottom of it when you have confidentiality issues?
I suggest you create a new condensed piece of software, stripped of all the proprietary stuff, and minimized for size, yet exibiting the problem. Probably best if it can (not) run on a Proto Board modified to 3.3 Volts, so we all have the same platform. If you post that, others can try it and confirm your findings, in which case I'm sure we can home in on the problem.
If we can't replicate your problem, yet you can with your hardware, then it is likely to be an issue with your particular set-up.
Truly, I have used almost a thousand chips in literally hundreds of designs, almost all at 50 Mhz, and many with very exacting requirements, and have (eventually) always been able to resolve any weird problems.
That is, except for one case years ago when we bought some apparently counterfeit chips from a broker (not from Parallax) that exhibited some similarities to what you describe.
Help us help you...... post some code that exhibits the problem, and we'll get this licked !! You are NOT alone here, this community will support you if you give them a chance.
Cheers,
Peter (pjv)
Post Edited (pjv) : 9/14/2007 11:51:08 PM GMT
The question I keep asking myself is:·How can voltage affect logic signals?··
The answer that keeps coming to mind, is that the SX is not 100% conditionally stable at 50Mhz and 3.3 volts.
If you've ever overclocked a PC, many times the CPU will seem stable. But·testing·with very specific
burn in routines·will show the CPU isn't stable. I just get the feeling that this is·a similar issue.
Now, has anyone ever tried to write code to really burn in an SX at a specific voltage/frequency?
If it's·avaliable, I'd like to get my hands one it.
Again, ask youself, why·will it work at 5 volts, and not 3.3 volts?
Thanks,
Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
guy started whining about the routing, so I moved a few pins. One pin landed on PortD with another pin also on D. It just happened that I had to modify a standard VP I had to mix a few features together and the changing of the pins needed to happen at the same time.·Of course I updated code and I·was using logical names for things and didn't make the connection.· What really threw me was most of my testing/development was with the SX at 5 volts, where everything worked·fine.·Crank it down to 3.3, and problem!
I guess this is what I get for taking on a job with custom packaging/circuit boards/RF comm·and·20 day delivery... [noparse]:([/noparse]
Oh well, good learning experience.
I guess what really bugs me, is that even with no loading on the pins (except high impedance scope lead) the chip acts
differently at 3 and 5 volts. I still think that's it's a bug. (Gunther would call it a feature, of course!·
Again, the assembler should really give us a warning about successive writes to a port. Just wondering, I don't use SXB, but
with that higher level language (SXB or C),·how to you know to watch for this sort of stuff?
Now, do we have a FAQ for bugs/Gotchas for this chip? Where these sorts of things are documented?
Anyway,
Thanks all!
Dan
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"A saint-like quantity of patience is a help, if this is unavailable, a salty vocabulary works nearly as well." - A. S. Weaver
-Phil
http://www.sxlist.com/techref/ubicom/sxports.htm already had a bit about RMW problems, but I've added the note about lower voltage operation there.
Please feel free to use the little form at the bottom of the page to add your comments as well.
(Who said the web site never gets updated? <grin>)
It's great to know that we can still do 3v ops at 50Mhz. Way to rock, Parallax!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
---
James Newton, Host of SXList.com
james at sxlist,com 1-619-652-0593 fax:1-208-279-8767
SX FAQ / Code / Tutorials / Documentation:
http://www.sxlist.com Pick faster!
Glad you found the problem.
This information is actually quite well documented in the Parallax literature, but I fear that many may not grasp it's significance.
In regard to high impedance scope leads, the standard 10 Meg probe has about 8 pF capacitance, and that's not insignificant.
For high speed work, I use an active probe with a 1GHz bandwidth GAS FET amplifier right in the tip, and it has only 1 pF load. What a difference those 7 pF make to the squareness and speed of an edge.
Cheers,
Peter (pjv)