So lets do a calculation to find out how well the resistors will or won't work.
First off for when the pin is pulled high using the circuit below
Rp
I2C bus--------+----------/\/\/\-----prop pin 3.6V
|
|---------/\/\/\--------5V
10k Ohm (I think that this is the recommended value
So the prop pin will be at about 3.6V because of the diode. So we have a (5-3.6)=1.4V drop to get to the prop pin. You are only supposed to let 0.5mA through the diode the wrong way according to the data sheet so your total resistance needs to be greater than
V=IR
R=V/I
R=1.4/0.0005=2800 Ohms which is fine in our case.
Now to check what happens when we try to pull the pin low
Rp
I2C bus--------+----------/\/\/\-----0V prop pin
|
|---------/\/\/\--------5V
10k Ohm (I think that this is the recommended value
Vbus=5V-10k*i
i=V/r=5/(10k+Rp)
Vbus=5-10k*5/(10k+Rp)
Lets pretend that the low voltage has to be less than 1 volt
1=5-50k/(10k+Rp)
4=50k/(10k+Rp)
40k+4Rp=50k
4Rp=10k
Rp=2.5k ohms
Now if we check the high voltage again
Vbus=5-10k*(5-3.6)/(10k+2.5k)
Vbus=5-14/12.5
Vbus=5-1.12
Vbus=3.88V
Which if I recall correctly was right on the limit. If we made Rp bigger it would raise the high voltage but it would also raise the lower voltage which also won't work.
steve,
i have changed my pullups from 3.3V (origionally planned for the 24LC256) to 5V.
now i have a series resistor of 1,47K between I2C and propeller and a 10K pullup to 5V. this should give me a low level of 0,64 volts and i high level of 3,52 volts, which is (well or just) within the specification for the 24C256, which is VILmax(0,3*Vcc)= 1,5 volts and VIHmin(0,7*Vcc)=3,5 volts.
However,if the prop is more than 10 minutes off, it wont start anymore; only after cycling the power switch again. I've attached some scope pictures of SCL and SDA line. While SCL looks ok, SDA seems a little bit strange to me.The pictures are not very good, but the EEPROM load process only takes appr. 1 second and the autofocus of my camera is relatively slow. Ignore that high line on the SDA picture; the shutter was open too long .
i will order a new prop today.
if i think of Jakacki's comments, that I2C is only sinking; how could i have fried my prop (saw the 3.3 power line going up to 4,2 volts)?
propelleruser999 said...
However,if the prop is more than 10 minutes off, it wont start anymore; only after cycling the power switch again.
I'm not quite following this. Is the prop not running for 10 minutes but still connected to power? How are you resetting it?
I would have made the resistor a bit bigger to get the voltages up a bit higher.
Remember that the prop is only rated to take a 3kV spike on the supply pins. This is fairly easy to generate just by walking around. Also, could you have shorted something out somewhere?
no it is disconnected via a power switch. there is no reset switch at all.
yes , i will try a 2K or 3K resistor, that fits better with the specs of the EEPROM chip.
i have displaced (just bent it out off the socket)pin 29 from the prop and it still shows 0.7 to 0.8 volts as low level( as with other pins,they are nearly 0 volts) ; but as you can see on my scope traces, it sometimes reaches 0 volts.
So the sequence to generate a failure is
1. Turn power to board off
2. Leave for 10 minutes
3. Turn power back on and prop won't start
4. If you quickly turn the power off and then on the prop starts fine?
It sure sounds like there could be something wrong with the pin. Don't know what caused it though. Wonder if it could be something related to the switch that you are using?
1. yes
2. yes
3. yes
4. mostly,but not always, its that grazy intermitting behaviour . once you got it running, you can cycle it 100 times within a few seconds and it always starts.
Did you see my SDA scope picture. What you see there , is a 3.3 spike, after that you see an appr. 0.8 volts signal. Is this data? then the level would be much too low.
i forgot to mention my scope settings
for both signals(SCL,SDA) it is y: 10us per division, x: 0,1 volts per division times 10.
Have you connected ALL prop Vss pins directly to GND and ALL prop VDD directly to +3.3v?·i.e connected externally - not relying on props internal connection - the prop could be damaged if (for example) you supplied one of the props vdd pins with 3.3v - then relying on the internal connection - connected another device to·another VDD pin·without an external link ... not directly associated with your issue but has been one of the root causes of PLL failures ..
I presume all reasonable precautions have been followed - common ground ? Decoupling Caps etc ..
Rgds,
John T
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ 'Necessity is the mother of invention'
Post Edited (QuattroRS4) : 4/14/2008 5:22:42 PM GMT
all 4 power pins are connected to either Vdd or Vss. All are decoupled with 0,33uf capacitors.
there is one rail for Vdd and one rail for Vss, directly connecting to the main Vss/Vdd rails via a short (1/2") wire.
in between and before voltage regulators there are 10 uF caps. were 100 uf before ( changed due to steves request)
- if starting with a dc power supply it always works. 100% if running from IDE to ram and EEPROM.
- if starting with a battery 9 volts (fully loaded)
it doesn't start if 2 minutes or longer off (but thats even not consistent,just had a 2 5 minute breaks and it started)
it always starts when recycling the power button several times within a few seconds and then forever if recycled within a few seconds
it mostly starts if touching one of the crystal pins with my finger
it always starts if the ground probe of my oscilloscope is connected to Vss
so its hard to find out, whether my clock is probably sometimes not starting. since only connecting my scope GND gets the prop running.
i several times changed the words always and mostly in the above text, since it is so varying.
1)so far i have not expreienced any problems, when running off a power supply
2) battery operation is very incalculable.this morning it started right off after a 10 hour sleep.10 minutes later it didnt satrt anymore. cycling the power switch a few times (power switch is not the problem,i have a led which shows me, that there is power) and it repeatedly worked,when cycled within a few seconds. Waiting another 10 minutes, i didnt start again.
the battery is a 9V NIMH 200 mah type. currently after a few 100 power cycles it shows 8,96 volts. fully loaded it was appr. 9,56 volts.
my vendor ELMICRO just told me, that new props are on the way to me.
1)so far i have not expreienced any problems, when running off a power supply
2) battery operation is very incalculable.this morning it started right off after a 10 hour sleep.10 minutes later it didnt satrt anymore. cycling the power switch a few times (power switch is not the problem,i have a led which shows me, that there is power) and it repeatedly worked,when cycled within a few seconds. Waiting another 10 minutes, it didnt start again.
the battery is a 9V NIMH 200 mah type. currently after a few 100 power cycles it shows 8,96 volts. fully loaded it was appr. 9,56 volts.
my vendor ELMICRO just told me, that new props are on the way to me.
Sounds like you are having power problems. Some NIMH batteries have 7 cells and others have 8. If yours has 8 than it is probably pretty flat (should be 9.6V). Try charging the battery and see if the problem goes away. I'd say that the 10 hours allows for the batteries chemistry to recharge and reduce the internal slightly which allows it to start. Than when you turn it off and back on you have taken the last bit of power from the battery and raised the resistance. Than with everything starting up and the 3 eeproms starting its asking for to much current for the poor battery to provide. A better solution might be to use 4 AA batteries.
i think i have an 8 cell NIMH battery,which accounts for the 9,58 volts, when fully loaded. My old ones only showed appr. 8.4 volts (probably 7 cells). they currently have 8.96 volts after a few hundred power on/off cycles. the symptoms havent changed between fully loaded and now. its nearly unpredictable, that the prop starts or not. only if once you get it started,and recycle it within a few seconds it always/almost starts again.
tomorrow or the day after tomorrow i'll get new props.
i still have concerns with my SDA line scope traces (short 3,3 volt spike,longer 0.5 to 0.8 voltage level,than long time 0 volts). i've also replaced the 24C256's (i have a bunch of them), they are not the cause. I've also isolated the SDA line from the prop (just bent the pin out) and it shows no really 0 volt level, all other i/o pins show a swing from 0 to 3,3 volts.
Also , as Jakicki said , that I2C devices are sinking only, not supplying; i do not understand, why my 3.3 volt rail went temporarely up to 4.2 volts.
today i received my new props. but that didnt change anything.
i have now added some additional caps (0,33uf) between Vdd and Vss directly to the EEPROM power pins. the caps on the prop are less than 1 inch away; but i feel, it has improved. no fails so far anymore. i have also changed parellel and serial resistors on the SDA line. I now have a voltage swing between 0.64 and 3,79 volts, which is now well within the specs of the 24C256/512 (Vlow max 0.3*Vdd = 1,5 V and Vhigh min 0.7*Vdd = 3,5 V).
I omitted caps on my 3 EEPROMs , because on the protoboard there are no ones shown in the schematic; but its may be that on the Protoboard it is only 1 EEPROM and it is mounted relatively near to the prop; and bigger ground and power planes, etc..
i started this post appr. 1 hour ago; i have cycled the power switch appr. 100 times and no boot fails so far.
with only parallel/serial resistors you cannot realize a 100 % reliable I2C communication between 3,3 and 5 volt devices. for reliable operation you need some active devices. If not risking the prop by using much lower resistor values to the 5 volts, you'll mostly fail on the 0,7 Vih*Vdd high min limit (3,5 volts),the Vlow max limit is no problem.
first fail after 300 power on cycles in between seconds, 10's of seconds , minutes, 10' s of minutes,hours; not bad.
Comments
First off for when the pin is pulled high using the circuit below
Rp I2C bus--------+----------/\/\/\-----prop pin 3.6V | |---------/\/\/\--------5V 10k Ohm (I think that this is the recommended valueSo the prop pin will be at about 3.6V because of the diode. So we have a (5-3.6)=1.4V drop to get to the prop pin. You are only supposed to let 0.5mA through the diode the wrong way according to the data sheet so your total resistance needs to be greater than
V=IR
R=V/I
R=1.4/0.0005=2800 Ohms which is fine in our case.
Now to check what happens when we try to pull the pin low
Rp I2C bus--------+----------/\/\/\-----0V prop pin | |---------/\/\/\--------5V 10k Ohm (I think that this is the recommended valueVbus=5V-10k*i
i=V/r=5/(10k+Rp)
Vbus=5-10k*5/(10k+Rp)
Lets pretend that the low voltage has to be less than 1 volt
1=5-50k/(10k+Rp)
4=50k/(10k+Rp)
40k+4Rp=50k
4Rp=10k
Rp=2.5k ohms
Now if we check the high voltage again
Vbus=5-10k*(5-3.6)/(10k+2.5k)
Vbus=5-14/12.5
Vbus=5-1.12
Vbus=3.88V
Which if I recall correctly was right on the limit. If we made Rp bigger it would raise the high voltage but it would also raise the lower voltage which also won't work.
So this looks like it won't work at all.
i have changed my pullups from 3.3V (origionally planned for the 24LC256) to 5V.
now i have a series resistor of 1,47K between I2C and propeller and a 10K pullup to 5V. this should give me a low level of 0,64 volts and i high level of 3,52 volts, which is (well or just) within the specification for the 24C256, which is VILmax(0,3*Vcc)= 1,5 volts and VIHmin(0,7*Vcc)=3,5 volts.
However,if the prop is more than 10 minutes off, it wont start anymore; only after cycling the power switch again. I've attached some scope pictures of SCL and SDA line. While SCL looks ok, SDA seems a little bit strange to me.The pictures are not very good, but the EEPROM load process only takes appr. 1 second and the autofocus of my camera is relatively slow. Ignore that high line on the SDA picture; the shutter was open too long .
i will order a new prop today.
if i think of Jakacki's comments, that I2C is only sinking; how could i have fried my prop (saw the 3.3 power line going up to 4,2 volts)?
I would have made the resistor a bit bigger to get the voltages up a bit higher.
Remember that the prop is only rated to take a 3kV spike on the supply pins. This is fairly easy to generate just by walking around. Also, could you have shorted something out somewhere?
no it is disconnected via a power switch. there is no reset switch at all.
yes , i will try a 2K or 3K resistor, that fits better with the specs of the EEPROM chip.
i have displaced (just bent it out off the socket)pin 29 from the prop and it still shows 0.7 to 0.8 volts as low level( as with other pins,they are nearly 0 volts) ; but as you can see on my scope traces, it sometimes reaches 0 volts.
1. Turn power to board off
2. Leave for 10 minutes
3. Turn power back on and prop won't start
4. If you quickly turn the power off and then on the prop starts fine?
It sure sounds like there could be something wrong with the pin. Don't know what caused it though. Wonder if it could be something related to the switch that you are using?
2. yes
3. yes
4. mostly,but not always, its that grazy intermitting behaviour . once you got it running, you can cycle it 100 times within a few seconds and it always starts.
Did you see my SDA scope picture. What you see there , is a 3.3 spike, after that you see an appr. 0.8 volts signal. Is this data? then the level would be much too low.
i forgot to mention my scope settings
for both signals(SCL,SDA) it is y: 10us per division, x: 0,1 volts per division times 10.
Have you connected ALL prop Vss pins directly to GND and ALL prop VDD directly to +3.3v?·i.e connected externally - not relying on props internal connection - the prop could be damaged if (for example) you supplied one of the props vdd pins with 3.3v - then relying on the internal connection - connected another device to·another VDD pin·without an external link ... not directly associated with your issue but has been one of the root causes of PLL failures ..
I presume all reasonable precautions have been followed - common ground ? Decoupling Caps etc ..
Rgds,
John T
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Necessity is the mother of invention'
Post Edited (QuattroRS4) : 4/14/2008 5:22:42 PM GMT
there is one rail for Vdd and one rail for Vss, directly connecting to the main Vss/Vdd rails via a short (1/2") wire.
in between and before voltage regulators there are 10 uF caps. were 100 uf before ( changed due to steves request)
- if starting with a dc power supply it always works. 100% if running from IDE to ram and EEPROM.
- if starting with a battery 9 volts (fully loaded)
it doesn't start if 2 minutes or longer off (but thats even not consistent,just had a 2 5 minute breaks and it started)
it always starts when recycling the power button several times within a few seconds and then forever if recycled within a few seconds
it mostly starts if touching one of the crystal pins with my finger
it always starts if the ground probe of my oscilloscope is connected to Vss
so its hard to find out, whether my clock is probably sometimes not starting. since only connecting my scope GND gets the prop running.
i several times changed the words always and mostly in the above text, since it is so varying.
better wait for my new props to arrive.
Is it fair to say ... ?
1) Works always and perfectly with a power supply.
2) Doesn't work well with battery.
what sort of battery ? PP3 / ~200mAh ?
1)so far i have not expreienced any problems, when running off a power supply
2) battery operation is very incalculable.this morning it started right off after a 10 hour sleep.10 minutes later it didnt satrt anymore. cycling the power switch a few times (power switch is not the problem,i have a led which shows me, that there is power) and it repeatedly worked,when cycled within a few seconds. Waiting another 10 minutes, i didnt start again.
the battery is a 9V NIMH 200 mah type. currently after a few 100 power cycles it shows 8,96 volts. fully loaded it was appr. 9,56 volts.
my vendor ELMICRO just told me, that new props are on the way to me.
1)so far i have not expreienced any problems, when running off a power supply
2) battery operation is very incalculable.this morning it started right off after a 10 hour sleep.10 minutes later it didnt satrt anymore. cycling the power switch a few times (power switch is not the problem,i have a led which shows me, that there is power) and it repeatedly worked,when cycled within a few seconds. Waiting another 10 minutes, it didnt start again.
the battery is a 9V NIMH 200 mah type. currently after a few 100 power cycles it shows 8,96 volts. fully loaded it was appr. 9,56 volts.
my vendor ELMICRO just told me, that new props are on the way to me.
Sounds like you are having power problems. Some NIMH batteries have 7 cells and others have 8. If yours has 8 than it is probably pretty flat (should be 9.6V). Try charging the battery and see if the problem goes away. I'd say that the 10 hours allows for the batteries chemistry to recharge and reduce the internal slightly which allows it to start. Than when you turn it off and back on you have taken the last bit of power from the battery and raised the resistance. Than with everything starting up and the 3 eeproms starting its asking for to much current for the poor battery to provide. A better solution might be to use 4 AA batteries.
Steven
i think i have an 8 cell NIMH battery,which accounts for the 9,58 volts, when fully loaded. My old ones only showed appr. 8.4 volts (probably 7 cells). they currently have 8.96 volts after a few hundred power on/off cycles. the symptoms havent changed between fully loaded and now. its nearly unpredictable, that the prop starts or not. only if once you get it started,and recycle it within a few seconds it always/almost starts again.
tomorrow or the day after tomorrow i'll get new props.
i still have concerns with my SDA line scope traces (short 3,3 volt spike,longer 0.5 to 0.8 voltage level,than long time 0 volts). i've also replaced the 24C256's (i have a bunch of them), they are not the cause. I've also isolated the SDA line from the prop (just bent the pin out) and it shows no really 0 volt level, all other i/o pins show a swing from 0 to 3,3 volts.
Also , as Jakicki said , that I2C devices are sinking only, not supplying; i do not understand, why my 3.3 volt rail went temporarely up to 4.2 volts.
i have now added some additional caps (0,33uf) between Vdd and Vss directly to the EEPROM power pins. the caps on the prop are less than 1 inch away; but i feel, it has improved. no fails so far anymore. i have also changed parellel and serial resistors on the SDA line. I now have a voltage swing between 0.64 and 3,79 volts, which is now well within the specs of the 24C256/512 (Vlow max 0.3*Vdd = 1,5 V and Vhigh min 0.7*Vdd = 3,5 V).
I omitted caps on my 3 EEPROMs , because on the protoboard there are no ones shown in the schematic; but its may be that on the Protoboard it is only 1 EEPROM and it is mounted relatively near to the prop; and bigger ground and power planes, etc..
i started this post appr. 1 hour ago; i have cycled the power switch appr. 100 times and no boot fails so far.
with only parallel/serial resistors you cannot realize a 100 % reliable I2C communication between 3,3 and 5 volt devices. for reliable operation you need some active devices. If not risking the prop by using much lower resistor values to the 5 volts, you'll mostly fail on the 0,7 Vih*Vdd high min limit (3,5 volts),the Vlow max limit is no problem.
first fail after 300 power on cycles in between seconds, 10's of seconds , minutes, 10' s of minutes,hours; not bad.