And if the device spends most of its time asleep, the marginal extra current won't matter much. However, my own opinion is that you don't need to change the crystal unless you need the smaller footprint. It should work the way it is. Maybe exchange the Xin and Xout traces,so the the one to Xin (pin 28 on the Prop) is the shorter of the two.
There are burst mode switching regulators that get down to much lower quiescent current, but nothing can beat the linear regulators with the pmosfet pass transistor in that department. It is really quite a design achievement by Microchip, although one does have to remain aware of its shortcomings. For heat dissipation, I really do recommend a low thermal resistance to the ground plane. The die itself will heat up very fast when the 50mA surge of current happens to the Prop+XBZB.
Your build technique with the laser and paint is very interesting! Oh, please do let us know on your earlier thread how the TDR sensor works out.
I have only skimmed this thread.
However, with a 9V input and 3v3 output, the regulator will be dissipating 5.7V @ 70mA would be 400mW. A SOT23 package without a sizable ground plane is not going to handle this IMHO (without doing proper temp calculations). Try supplying 6V and see how that goes.
One other point I noticed is that your 3V3 power to the prop Vcc pins is not close at all - they are routed around the boards ground plane. Both the GND and VCC pins all need to be connected close together (and you must connect all 4 sets of them in the QFP package).
Try your circuit without the xtal being set in your program. You may have just damaged the PLL in the prop, in which case everything else may function except the xtal oscillator.
Oh, please do let us know on your earlier thread how the TDR sensor works out.
Will do sir!
I have already incorporated most of the changes discussed here and I hope to have the new board ready for etching by Monday. Just going to swap around the crystal leads and will be good to go. Thanks for all the help!!
Now I see your reason for using 9V, because you are using batteries and a 9V is about the same size as 2 AA cells. But 2 AA cells have much greater capacity than a 9V and are much cheaper. There are numerous step-up regulator chips available for just this purpose where you can step-up a single or dual cell up to a nicely regulated 3.3V and I think that if you do put the regulator in standby that you will still get close to the full battery voltage so that means if you disable brown-out and switch to the internal oscillator you can run on this reduced voltage during standby while sipping mere electrons.
Now I see your reason for using 9V, because you are using batteries and a 9V is about the same size as 2 AA cells. But 2 AA cells have much greater capacity than a 9V and are much cheaper. There are numerous step-up regulator chips available for just this purpose where you can step-up a single or dual cell up to a nicely regulated 3.3V and I think that if you do put the regulator in standby that you will still get close to the full battery voltage so that means if you disable brown-out and switch to the internal oscillator you can run on this reduced voltage during standby while sipping mere electrons.
Yep that is actually plan C. If I cant get this configuration to work to my satisfaction I'm going to go back to 3 AA batteries and see if I can find a case I like. If cant find a case I'm going this route. If that is what I end up doing you can bet I will be back looking for advice. I have only played with one circuit that stepped-up the voltage and it was a combo circuit that did both step-up and down and was not efficient in either direction.
I like the idea of turning off the regulator and just running from the two AA batteries while sleeping. Hmm that's got me scratching my chin!!
If you do attempt a low voltage design pleas read through the thread I started a while ago. Prop-limbo!
TLDR, with the brownout disabled rcslow and rcfast work fine down to 1.5v but both clocks are likely to stop below 1.4v. (should be able to go lower with an external oscillator) Didn't test the crystal oscillator or PLL.
Quick update:
Yesterday my parts came in for my new board and everything works great!! I made several small modifications to my PCB layout but I think the modifications that helped the most are the 220uF caps to the input and output of the MCP1701A as well as the change to a SOT-89 package with a big copper pour under its thermal tab instead of the smaller SOT23 (thanks again Tracy for that recommendation). This time I built out the power circuit on the PCB first so I could test before installing the Propeller. I hooked up my scope to the output and watched the 3.3v for any oscillation or power surges. While watching the scope I connected and disconnected a 330 ohm resistor to simulate a 100mA load. Everything looked great. For the thermal test I hooked up two 330 ohm resistors in parallel with an LED that pushed the load up to 215mA. I then ran with that load for 15 minutes and the scope showed a solid 3.3 volts DC (no AC ripple) solid as a rock. I used my iPhone's FLIR thermal camera (very cool Christmas present my wife gave me last year) to get the temperature of the MCP1702T-3302E/MB and it was around 82C (180F) well below the 125C degree Max operating temperature. It was very warm to the touch (I was able to keep my finger on it to a count of 30). That is over twice the max current and 15 times longer than what my circuit requires. My circuit pulls less than 100mA (~70mA) for about 45 seconds once an hour. The rest of the time the load is only 40uA (Prop is running on internal slow clock) with small 10 microsecond burst to 1mA every 10 seconds as the xBee checks for data.
I think I'm going to be happy with this solution!! Thanks everyone for the help!!
Quick update:
Yesterday my parts came in for my new board and everything works great!! I made several small modifications to my PCB layout but I think the modifications that helped the most are the 220uF caps to the input and output of the MCP1701A as well as the change to a SOT-89 package with a big copper pour under its thermal tab instead of the smaller SOT23 (thanks again Tracy for that recommendation). This time I built out the power circuit on the PCB first so I could test before installing the Propeller. I hooked up my scope to the output and watched the 3.3v for any oscillation or power surges. While watching the scope I connected and disconnected a 330 ohm resistor to simulate a 100mA load. Everything looked great. For the thermal test I hooked up two 330 ohm resistors in parallel with an LED that pushed the load up to 215mA. I then ran with that load for 15 minutes and the scope showed a solid 3.3 volts DC (no AC ripple) solid as a rock. I used my iPhone's FLIR thermal camera (very cool Christmas present my wife gave me last year) to get the temperature of the MCP1702T-3302E/MB and it was around 82C (180F) well below the 125C degree Max operating temperature. It was very warm to the touch (I was able to keep my finger on it to a count of 30). That is over twice the max current and 15 times longer than what my circuit requires. My circuit pulls less than 100mA (~70mA) for about 45 seconds once an hour. The rest of the time the load is only 40uA (Prop is running on internal slow clock) with small 10 microsecond burst to 1mA every 10 seconds as the xBee checks for data.
I think I'm going to be happy with this solution!! Thanks everyone for the help!!
Glad to hear that you got it working. It's just that when I hear things like "220uF" on the output of the regulator I cringe. Either a regulator regulates or it doesn't, but of course it does. If the regulator is a little slow it may need a few microfarads, but hundreds and sometimes I even hear thousands, why, that's a huge overkill. I'd be surprised if you needed more than 10uF with an MCP1702. Having a lot of capacitance on the input of the regulator might be necessary depending upon the load and the distance from the power source.
Of course you know too that 82'C "heat" is all that precious energy going to waste from an already inefficient power source, the common 9V battery. I'd like to see you implement plan C (C for Cool), boosting the two AA cells up to working voltage and then putting the regulator into standby along with the Prop.
Glad to hear that you got it working. It's just that when I hear things like "220uF" on the output of the regulator I cringe. Either a regulator regulates or it doesn't, but of course it does. If the regulator is a little slow it may need a few microfarads, but hundreds and sometimes I even hear thousands, why, that's a huge overkill. I'd be surprised if you needed more than 10uF with an MCP1702. Having a lot of capacitance on the input of the regulator might be necessary depending upon the load and the distance from the power source.
Of course you know too that 82'C "heat" is all that precious energy going to waste from an already inefficient power source, the common 9V battery. I'd like to see you implement plan C (C for Cool), boosting the two AA cells up to working voltage and then putting the regulator into standby along with the Prop.
I'm with you on the cap for the output side I think I will dial it down a little those things are big and I need the space in my next version.
On plan C (Cool) are you thinking about feeding the prop's power from the voltage boost circuit and then for sleep mode shut the voltage boost off and continue to pull the 40 to 50uA from the boost? In this case a big cap on the output side would be a good idea. I may have a problem with this configuration as my xBee is also in sleep mode but it does wake up every 10 seconds and polls its parent for data. During the poll its power draw will jump up to just under 1mA for 10 milliseconds. If the xBee's parent has queued data for it the xBee will start to receive the data (jumps to 10mA) and at the same time pull a pin low waking up the propeller. It is as at this time the propeller would know to switch the power boost circuit back on and I fear that would be too late, the circuit would have exhausted the power in the cap.
Here is how I was thinking plan C would work: There is only one component that has to have an exact 3.3v supplied to it, the Atlas Scientific TDS circuit I'm using. In today's design I switch the TDS circuit on and off with an external BJT to save power. Since the xBee, Prop, and EEPROM can get by on voltages from 3.6v down to 2.1 (xBee's low limit) I was thinking about connecting them directly to the two batteries. I would then replace the BJT transistor with the 3.3v boost circuit and switch it on when I need to take a reading. Basically I would only use the voltage boost circuit to drive the Atlas Scientific TDS circuit. The two AA batteries would connect directly to a PMOS FET (for polarity reversal protection) and then to a nice big 220uF cap. From that cap I would directly feed the Propeller, xBee, EEPROM, and the voltage boost circuit.
Thoughts? Am I asking for trouble going this route?? Do you have a voltage boost circuit / component you would recommend? I would like to read through its data sheet.
John, One such boost chip I've used is the LT1615, which comes in a simple SOT23-5 package. It is not the most efficient, but it is easy to lay out on a pcb for the kind of intermittent usage you are contemplating.
The diagram is for a boost from an LiPoly single cell to 7V, but it is also suitable with choice of feedback for the boost from 2*AA up to 3.3V. Be aware that the shutdown pin does not shut off the output voltage. It only shuts off the boost. That is due to the DC path through the inductor and diode, and it might not be what you want for the scheme where you power only the TDS from the booster. To shut off the output entirely, you have to add series switch for the whole circuit, or go to a SEPIC configuration, both more complicated.
I think what Peter was talking about was to have the entire circuit powered from the output of an efficient booster. Turn on the boost only when necessary to operate the TDS. Otherwise the main circuit operates from the pass-through voltage.
I think what Peter was talking about was to have the entire circuit powered from the output of an efficient booster. Turn on the boost only when necessary to operate the TDS. Otherwise the main circuit operates from the pass-through voltage.
Thanks Tracy! I bet that is exactly what Peter is talking about. I didn't know the "un-boosted" voltage would be available if the booster was off.
My new circuit (based on the 9v battery) is just flat out rocking!! Yea it got a little warm during my stress testing but now that I have been running for a few days I have not noticed it heating up at all and it is darn reliable and efficient. It also has cleaned up some noise that was causing retransmissions of ZigBee packets! I think the big output cap is helping that a lot!! At this point the only reason I would go the boost route is if the 800mAh I can get out of a 9V battery is not going to be enough. Right now its not, my estimated run time today is 5 to 6 months and I would like to get a year. But I have just started to fine tune my code and the radio. The TDS circuit is a power hog and they are working on an update that I hope will be more efficient.
Comments
In this discussion http://forums.parallax.com/showthread.php/103634-Spin-Stamp-Crystal-10MHz-with-PLL on post 5 Phil talks about how this crystal will create an internal base frequency of 160MHz before it is divided down to 80MHz. That sounds like more power.
https://www.parallax.com/product/28327
https://www.parallax.com/sites/default/files/downloads/28327-PropellerBackpack-v1.0.pdf
EDIT: He is logged on, so he may chime in if he reads this.
There are burst mode switching regulators that get down to much lower quiescent current, but nothing can beat the linear regulators with the pmosfet pass transistor in that department. It is really quite a design achievement by Microchip, although one does have to remain aware of its shortcomings. For heat dissipation, I really do recommend a low thermal resistance to the ground plane. The die itself will heat up very fast when the 50mA surge of current happens to the Prop+XBZB.
Your build technique with the laser and paint is very interesting! Oh, please do let us know on your earlier thread how the TDR sensor works out.
However, with a 9V input and 3v3 output, the regulator will be dissipating 5.7V @ 70mA would be 400mW. A SOT23 package without a sizable ground plane is not going to handle this IMHO (without doing proper temp calculations). Try supplying 6V and see how that goes.
One other point I noticed is that your 3V3 power to the prop Vcc pins is not close at all - they are routed around the boards ground plane. Both the GND and VCC pins all need to be connected close together (and you must connect all 4 sets of them in the QFP package).
Try your circuit without the xtal being set in your program. You may have just damaged the PLL in the prop, in which case everything else may function except the xtal oscillator.
Thanks Cluso for the input. I think I'm in pretty good shape now.
Will do sir!
I have already incorporated most of the changes discussed here and I hope to have the new board ready for etching by Monday. Just going to swap around the crystal leads and will be good to go. Thanks for all the help!!
I like the idea of turning off the regulator and just running from the two AA batteries while sleeping. Hmm that's got me scratching my chin!!
Thanks Peter for the help!!
TLDR, with the brownout disabled rcslow and rcfast work fine down to 1.5v but both clocks are likely to stop below 1.4v. (should be able to go lower with an external oscillator) Didn't test the crystal oscillator or PLL.
Marty
Yesterday my parts came in for my new board and everything works great!! I made several small modifications to my PCB layout but I think the modifications that helped the most are the 220uF caps to the input and output of the MCP1701A as well as the change to a SOT-89 package with a big copper pour under its thermal tab instead of the smaller SOT23 (thanks again Tracy for that recommendation). This time I built out the power circuit on the PCB first so I could test before installing the Propeller. I hooked up my scope to the output and watched the 3.3v for any oscillation or power surges. While watching the scope I connected and disconnected a 330 ohm resistor to simulate a 100mA load. Everything looked great. For the thermal test I hooked up two 330 ohm resistors in parallel with an LED that pushed the load up to 215mA. I then ran with that load for 15 minutes and the scope showed a solid 3.3 volts DC (no AC ripple) solid as a rock. I used my iPhone's FLIR thermal camera (very cool Christmas present my wife gave me last year) to get the temperature of the MCP1702T-3302E/MB and it was around 82C (180F) well below the 125C degree Max operating temperature. It was very warm to the touch (I was able to keep my finger on it to a count of 30). That is over twice the max current and 15 times longer than what my circuit requires. My circuit pulls less than 100mA (~70mA) for about 45 seconds once an hour. The rest of the time the load is only 40uA (Prop is running on internal slow clock) with small 10 microsecond burst to 1mA every 10 seconds as the xBee checks for data.
I think I'm going to be happy with this solution!! Thanks everyone for the help!!
Glad to hear that you got it working. It's just that when I hear things like "220uF" on the output of the regulator I cringe. Either a regulator regulates or it doesn't, but of course it does. If the regulator is a little slow it may need a few microfarads, but hundreds and sometimes I even hear thousands, why, that's a huge overkill. I'd be surprised if you needed more than 10uF with an MCP1702. Having a lot of capacitance on the input of the regulator might be necessary depending upon the load and the distance from the power source.
Of course you know too that 82'C "heat" is all that precious energy going to waste from an already inefficient power source, the common 9V battery. I'd like to see you implement plan C (C for Cool), boosting the two AA cells up to working voltage and then putting the regulator into standby along with the Prop.
I'm with you on the cap for the output side I think I will dial it down a little those things are big and I need the space in my next version.
On plan C (Cool) are you thinking about feeding the prop's power from the voltage boost circuit and then for sleep mode shut the voltage boost off and continue to pull the 40 to 50uA from the boost? In this case a big cap on the output side would be a good idea. I may have a problem with this configuration as my xBee is also in sleep mode but it does wake up every 10 seconds and polls its parent for data. During the poll its power draw will jump up to just under 1mA for 10 milliseconds. If the xBee's parent has queued data for it the xBee will start to receive the data (jumps to 10mA) and at the same time pull a pin low waking up the propeller. It is as at this time the propeller would know to switch the power boost circuit back on and I fear that would be too late, the circuit would have exhausted the power in the cap.
Here is how I was thinking plan C would work: There is only one component that has to have an exact 3.3v supplied to it, the Atlas Scientific TDS circuit I'm using. In today's design I switch the TDS circuit on and off with an external BJT to save power. Since the xBee, Prop, and EEPROM can get by on voltages from 3.6v down to 2.1 (xBee's low limit) I was thinking about connecting them directly to the two batteries. I would then replace the BJT transistor with the 3.3v boost circuit and switch it on when I need to take a reading. Basically I would only use the voltage boost circuit to drive the Atlas Scientific TDS circuit. The two AA batteries would connect directly to a PMOS FET (for polarity reversal protection) and then to a nice big 220uF cap. From that cap I would directly feed the Propeller, xBee, EEPROM, and the voltage boost circuit.
Thoughts? Am I asking for trouble going this route?? Do you have a voltage boost circuit / component you would recommend? I would like to read through its data sheet.
Thanks Peter!!
The diagram is for a boost from an LiPoly single cell to 7V, but it is also suitable with choice of feedback for the boost from 2*AA up to 3.3V. Be aware that the shutdown pin does not shut off the output voltage. It only shuts off the boost. That is due to the DC path through the inductor and diode, and it might not be what you want for the scheme where you power only the TDS from the booster. To shut off the output entirely, you have to add series switch for the whole circuit, or go to a SEPIC configuration, both more complicated.
I think what Peter was talking about was to have the entire circuit powered from the output of an efficient booster. Turn on the boost only when necessary to operate the TDS. Otherwise the main circuit operates from the pass-through voltage.
Thanks Tracy! I bet that is exactly what Peter is talking about. I didn't know the "un-boosted" voltage would be available if the booster was off.
My new circuit (based on the 9v battery) is just flat out rocking!! Yea it got a little warm during my stress testing but now that I have been running for a few days I have not noticed it heating up at all and it is darn reliable and efficient. It also has cleaned up some noise that was causing retransmissions of ZigBee packets! I think the big output cap is helping that a lot!! At this point the only reason I would go the boost route is if the 800mAh I can get out of a 9V battery is not going to be enough. Right now its not, my estimated run time today is 5 to 6 months and I would like to get a year. But I have just started to fine tune my code and the radio. The TDS circuit is a power hog and they are working on an update that I hope will be more efficient.