the best power regulator

[John Gutmann]

what is the best power regulator for 5 volts to use with TTL? I know there are some good ones in the TO-220 package but idk what ones. any ideas? links?

[Matt Krass]

Regulating down from what? You might want to check out Maxim ICs (www.maxim-ic.com) level shifters, nice stuff and they'll usually give you free samples.

[Mike Betts]

what is the best power regulator for 5 volts to use with TTL? I know there are some good ones in the TO-220 package but idk what ones. any ideas? links?

Without more information as to exactly what your specifications are, the 7805 is the industry standard for TTL applications.

The 7805 is made by many manufacturers. Here is National's implementation (http://www.national.com/pf/LM/LM78M05.html).

Mike

[Hutch]

Ugh, sorry about the Puzzler post... Definitely left logged in and I didn't notice.

Anyhow, I'd say the best solution would be the regulator but if you don't have one, another (possibly cheaper) solution could be made with appropriately rated zener diodes.

EDIT: Found some decent diagrams and an explanation of how to do things the zener diode way, as well as how to do it with regulators (just keep hitting next slide):

http://www.mines.edu/Academic/courses/physics/phgn217/lab7/lect11/sld005.htm

[John Gutmann]

down from 12 volts

[John Gutmann]

so if i used a 12 volt in put and get +5volts as an output then what do i use as ground?

[BrianBSL]

so if i used a 12 volt in put and get +5volts as an output then what do i use as ground?

On a 7805, and most regulators, the grounds are tied togeather. on a TO-220 package, the metal tab is usually internally connected to the ground as well (so make sure you don't short anything out on it).

[Hutch]

Yep voltage is all relative, so it's output 5 volts relative to the wire the chip's ground is attached to.

[John Gutmann]

so i use the output ground as the input ground, and i have seen schematics of 7805's that hav non-polarized capacitors between the leads and i was wondering if those are neccessary, if they are what size should I use? and if not should i use somthing else

[Hutch]

They aren't completely necessary, but they will prevent current spikes. When logic gates change states, they draw a lot of current (relatively), and the capacitors across the leads will keep the the current drawn from spiking as much.

[ConKbot of Doom]

Yes, the caps are necessary. The stabilize the regulator and keep it from oscillating. Larger caps however, will help out with current spikes. Larger caps in addition to the small ones, not instead of.

[Hutch]

Hmmm I didn't recall them being necessary. But I at least am 99% sure that you can have the large ones instead of small ones since, after all, parallel capacitors are like one bigger capacitor (hooray for narrow, worthless AP physics knowledge :D ).

[John Gutmann]

so for the ground i use, do i use the same ground as what i use for the transister or the one on the part on the back?

what size caps? should it be greater than a certain size or will the supply just be more stable if i use bigger ones? what is the max size i should use? any recommendations on sizes?

and are there any other transistors that can be used the same way, i think maybe the 7804, the 7803 and 7806, i am not sure so someone please correct me

also should i use a small heat sink with it? it is required or should i used one just for insurance?


thx
-john-

[Joe Ross]

The exact size of the capacitors is absolutely necessary for ensuring stability (unless you really know what you are doing). Unfortunately, the real world doesn't always work like AP Physics. The datasheet for your regulator will tell you what size capacitor and any other requirements like type (tantalum, ceramic, etc), locations, etc *.

The datasheet for the 7805/LM340 is available here: http://www.national.com/ds.cgi/LM/LM340.pdf. If the part you get is not manufactured by national, you should get the datasheet from that manufacturer. There may be a few subtle differences (not likely for the 7805, but you'll probably run across a difference in some other chip before too long).

As far as heatsinks, it depends on how much current you plan to draw and the package of the regulator. Looking at the national datasheet, at 25 degrees C, the TO-220 can dissipate about 2 watts, although for safety, lets use 1 watt. The equation for power dissipation can be approximated by Pdis~=(Vin-Vout)*Iin. Since Vin is 12v and vout is 5v, you can solve for Iin of 1/7 of an amp. Since that's Iin, your output current (Iout) could actually be 350ma. I have a feeling that's more then enough for your application.

All of the grounds should be hooked together, the ground from your source, the ground from your regulator (middle pin of the TO-220 as well as the case), and the ground of whatever the load or loads are (IE your TTL chips and anything else).

Note that I've never designed a regulator circuit (other then a bridge rectifier), so you'll be a step ahead of me. I have worked with a bunch, though.

* I've seen a person with an EE degree and several years industry experience that designed a regulator circuit that didn't quite work because he didn't follow the requirements listed in the datasheet.

[Manoel]

As far as heatsinks, it depends on how much current you plan to draw and the package of the regulator. Looking at the national datasheet, at 25 degrees C, the TO-220 can dissipate about 2 watts, although for safety, lets use 1 watt. The equation for power dissipation can be approximated by Pdis~=(Vin-Vout)*Iin. Since Vin is 12v and vout is 5v, you can solve for Iin of 1/7 of an amp. Since that's Iin, your output current (Iout) could actually be 350ma. I have a feeling that's more then enough for your application.

Joe,

Voltage regulators are not transformers, unfortunately. The regulation is obtained by means of power dissipation, as you pointed out. I agree I_{in} (is this LaTeX? :p) is 1/7 of an amp, but Iout must be essentially the same, minus some small leakage to ground. To make 1/7 of an amp at the input become 350 mA at the output you considered that the input power is equal to the output power, as in an ideal transformer (Vin*Iin = Vout*Iout), but that contradicts your first statement about power dissipation and thus, Iout is not 350 mA.
1/7 of an amp is approximately 142 mA, which is still a good number for sparksandtabs's use (digital logic circuits).

[John Gutmann]

but for Iout wont it only be equal to the Iin of what ever if hooked to the output of the 7805, so the Iin would be equal to the Iin of the 7805 + the Iin of what ever is on the other side?

[Joe Ross]

but for Iout wont it only be equal to the Iin of what ever if hooked to the output of the 7805, so the Iin would be equal to the Iin of the 7805 + the Iin of what ever is on the other side?

Manoel is correct about Iin and Iout being equal. You have a good point that they aren't exactly equal however. Iin of the 7805 is usually small compared to the load. I've seen people use the factor of Iin=1.01Iout (although that was for more efficient linear regulators then a 7805). I'd use something higher for safety, but it doesn't matter if your Iout is low.

[Hutch]

The exact size of the capacitors is absolutely necessary for ensuring stability (unless you really know what you are doing). Unfortunately, the real world doesn't always work like AP Physics. The datasheet for your regulator will tell you what size capacitor and any other requirements like type (tantalum, ceramic, etc), locations, etc *.



The data sheet you gave claimed the output capacitor wasn't necessary for stability, only transients... So now I'm a bit confused... Perhaps you were refering to the input capacitors? But even then the data sheet says they are only necessary "far from the power supply". :confused:

NOTE: I'm not a EE, I'm not trying to be smart, I just want to be educated if I'm wrong :).

[KenWittlief]

the regulator works by acting like a variable resistor. Depending on how much current your load is taking, it changes its resistance to keep the voltage steady.

So depending on how fast it can respond to changes in the load, it might become unstable unless you put the caps on both sides.

If the input is located within inches of a battery or a filtered output from a transformer and rectifier, then you might not need the input cap

but we are talking about what? 5 cents? why not put the caps on both sides, build the thing once and you are done with it.

[Hutch]

Well the issue was more of: I don't understand why you would need seperate (parallel) capacitors for filtering transients and stability... Especially when the data sheet claims you don't need them for stability. I'm just a little confused :confused:

[Gdeaver]

The default or standard caps for a 7805 are a 10uf aluminum electrolytic cap on both the in and out. Could you get by with out them? Depends what the regulator is driving. A pure resistive load like a light bulb wouldn't need them. Any digital electronics need a well filtered power supply. I always use at least a 100 uf on the output.

[Hutch]

I know that... I said they were needed for transients (generated by digital logic and that). But what I don't get is having seperate capacitors for stability.

[Al Skierkiewicz]

Guys,
I have to step in here and add a few things. As to the 78XX series of regulators, they are very sophisticated devices and include temperature compensation, voltage feedback and over current protection circuits. In power supply design using this series, keep leads on everything short so that series resistance and inductance don't cause variations in the output. The negative side of all capacitors, the regulator and the power common (the negative side of the bridge rectifier or transformer center tap if part of an AC supply) should all be tied together at one point. Since the 78XX series are active devices they can oscillate if not handled properly. Although your AP physics has taught you that capacitors in parallel add together, this is one of the cases where that is not exactly true. Different capacitor types act differently at high frequency than they do at low frequency. As such, use the cap type specified in the data sheet. It is not uncommon to find a large electrolytic in parallel with a small ceramic cap. The electrolytic supplies current to heavy loads and acts as a low impedance source at low frequencies while the small ceramic cap prevents high frequency oscillations and prevents phase shifts from existing from input to output. As a power device, a 7805 can become a very effective transmitter producing a 5 volt square wave of RF energy up into the 100kHz-1 MHz range that will drive you digital circuitry nuts. A big problem for designers using a 7805 from a 12 volts source is getting rid of heat. In the previous examples, drawing 350 ma, the device is being asked to dissipate 2.45 watts minimum. Without a heatsink, the little package gets very hot and the protection circuits cut in to protect the device. When breadboarding your design be careful around the tab of the case, it can get hot enough to burn the tender parts of your hand. Many designers will use the case of their project or the chassis as a heatsink.

[KenWittlief]

Al speaks of power supply filtering black magic :^)

the reason why real caps 'dont exactly add in parallel' is because every cap has some series resistance and series inductance.

At low frequencies these second order parameters have no effect. But at higher frequences they become significant.

This is why a small value cap performs better at high frequencies, because it is physically smaller than, lets say, a big old 1000µF 16V electrolytic can.

A small cap will have much lower series resistance and inductance.

Hope this makes more sense now.

[Hutch]

Ah ok, that confirms what I read yesterday when I decided to just go find out. Thanks :).

EDIT: And sorry for kind of derailing the thread :o.

[Gdeaver]

The cap issue is of minor concern, the real problem is the heat that running on a 12 volt supply will generate. What is the regulator driving. How many ma? Is this for a first robot project or something else? There are many advantages to using a separate 7.2 or 8.4 battery pack for the electronics and let the 12 volts drive the motors. For a First project, the PWM ports can supply 7.2 volts to an addition electronics board. This comes from the nicad battery that is plugged into the robot controller. The cmu cam was supplied this way. This power has to be shared with all devices plugged into the PWM outputs. The details are in the robot manual.

[John Gutmann]

but how ma can the pwms supply?

[Gdeaver]

Ops, I thought I saw this in the robot manual. The Analog power pins can provide a total of 1 amp 5 volts regulated. It is fused. First recommends a limit of 50 ma per pin. This is in the first 2005 robot manual. The only place I found a reference to the power available to the PWM pins is in the IFI question and answer for the RC. A total of 4 amps at 7.2 volts is available on the PWM pins which is supplied by the backup battery. There is no mention if it is fused. The victors do not draw from the pwm pins. I don't know about spikes. I believe I read that the radio modem is powered by the backup battery. The CMU camera uses this power but I didn't find a total amp draw for the camera and boards. I also couldn't find an amp dram at 7.2 volts for the hitec servos. Still there would seem to be plenty of power at 7.2 volts available for most additional electronics. A 5 volt regulator would be allot happier if supplied by the PWM pins instead of the 12 volt main feed. One caution- there should be only 1 ground path. On the CMU the ground is threw the rs232 port.

[John Gutmann]

who sed i am using it with the robot controller?

i want to use them for my custom robot i have built

[Al Skierkiewicz]

Ops, I thought I saw this in the robot manual. The Analog power pins can provide a total of 1 amp 5 volts regulated. It is fused. First recommends a limit of 50 ma per pin. This is in the first 2005 robot manual. The only place I found a reference to the power available to the PWM pins is in the IFI question and answer for the RC. A total of 4 amps at 7.2 volts is available on the PWM pins which is supplied by the backup battery. There is no mention if it is fused. The victors do not draw from the pwm pins. I don't know about spikes. I believe I read that the radio modem is powered by the backup battery. The CMU camera uses this power but I didn't find a total amp draw for the camera and boards. I also couldn't find an amp dram at 7.2 volts for the hitec servos. Still there would seem to be plenty of power at 7.2 volts available for most additional electronics. A 5 volt regulator would be allot happier if supplied by the PWM pins instead of the 12 volt main feed. One caution- there should be only 1 ground path. On the CMU the ground is threw the rs232 port.

Before things get too far here, a 78XX series regulator needs at least 2 volts across the regulator to operate as a stable voltage source. (this varies as a function of load current) Using a 7.2 volt battery as the input will not give you the regulation you might need for your custom circuit. When the input drops below the 2 volt differential, the output goes down as well. (usually in bad and unpredictable ways.) Although the backup battery on the RC does supply power to the servo outputs, it is also supplying a variety of other things like the main processor, radio modem and other keep alive circuitry ( and this year the CMU and all of it's related servoes, about 250 ma max.) so I don't suggest using that power on an FRC robot using an RC. In addition, the robot electrical manual requires that custom circuits be fed through their own circuit breaker off the 12 volt buss.

Since this application does not use the RC all of that is mute. The cap issue is not a minor concern when using digital custom circuits. The capacitor recomendations are part of the output impedance set point for the regulator. As the frequency increase so does the output impedance. The caps stabilize the output impedance and prevent the regulator from being influenced by rapidly changing loads. Depending on the type of digital circuits you are using (TTL, CMOS) they have different current usage parameters and anything that affects the current provided affects the switching points and the outputs. As Ken said above, we are talking some black magic kind of electronics at this point, follow the data sheets.

Finally, both the Victors and the Spikes draw primary operating power from the power input pins, but the control signals (PWM and relay control) are supplied by the 5 volt regulator on board the RC.

[Gdeaver]

This thread hasn't really addressed the original posted question and it has a important direct bearing on First and next year's game. There has been allot of posts on CD this summer concerning coprocessors. Seems allot of teams are looking to add coprocessors to manage additional sensors. They are going to need power from the robot.
Is there a better regulator than the 7805 series for 12 volts that is easy to use and readily available? In other words can a reference design be suggested that world work well and avoid most pitfalls.
Since it seem that there maybe allot more teams actually using additional electronics, are the advantages to using the 7.2 volts from the back up battery worth having First consider opening up that source for additional electronic? This also applies to the use of a separate battery for electronics that the original poster may consider using. Is it worth it? As Al pointed out the use of a non LDO regulator has some problems. Then that brings up the subject of is there a good reference design for using a LDO regulator for the 7.2 volt supply? I know Power supply design is a complex topic and the details can be debated around and around. May be those with the knowledge and experience could jump into this thread and help with some reference designs and guide lines to avoid pitfalls.

[KenWittlief]

To give better power supply design details, the requirements need to be specified:

1. how much current is required on the 5V line?

2. how much ripple or noise is acceptable?

3. is effeciency important for the application?

4. how much room is available for the regulator?

5. must the design be limited to through-hole componets or can the design implement surface mount parts (which you might not be able to solder by hand)?

6. are you making a printed circuit board or are you going to assemble the parts on a proto-board?

there is no one-size-fits-all 'best' voltage regulator circuit.

[Al Skierkiewicz]

Add to Ken's list...
reserve power needs, input voltage (some devices can provide 5 volts from a single "D" cell) effects from stray magnetic fields (some use multi winding coils and high frequency oscillators) and location for cooling if needed.

As Gdeaver has stated LDO (Low Drop Out) voltage regulators are available. If money is no object then anything is possible. However, using a good 7.2 volt source and a couple of diodes get you down to almost 5 volts too, 3 x 0.6 volts = 1.8 volts with no current drops. If the current load always stays the same a current source can be assembled using a FET and resistor

[John Gutmann]

To give better power supply design details, the requirements need to be specified:

1. how much current is required on the 5V line?

2. how much ripple or noise is acceptable?

3. is effeciency important for the application?

4. how much room is available for the regulator?

5. must the design be limited to through-hole componets or can the design implement surface mount parts (which you might not be able to solder by hand)?

6. are you making a printed circuit board or are you going to assemble the parts on a proto-board?

there is no one-size-fits-all 'best' voltage regulator circuit.

1. less then 200mA

2. as little noise as possible i will just be using normal TTL ICs

3. as efficient as possible, but effeciency for what? voltage?

4.as much room as as needed

5. through hole components

6. probaly a bread-board to test it then finalize it on a printed circuit board

[John Gutmann]

also I AM NOT GOING TO USE IT WITH THE RC IF I DO, IT WILL BE WITH AN OLDER PBASIC ONE WHICH RUNS STRAIGHT FROM 12 VOLT BATTERY SO PLEASE STOP SAYING I CAN DO IT WITH THE RC!!!!!!!!!!!!!!!

[BrianBSL]

also I AM NOT GOING TO USE IT WITH THE RC IF I DO, IT WILL BE WITH AN OLDER PBASIC ONE WHICH RUNS STRAIGHT FROM 12 VOLT BATTERY SO PLEASE STOP SAYING I CAN DO IT WITH THE RC!!!!!!!!!!!!!!!

You know, a lot of people have put time in responding to your request that was not very well outlined, and you show very little appreciation to them despite the fact that they took time to respond to your post. I've noticed this is not the only post of yours that you have given a response like this. This forum is about FIRST robotics, not your special robot, so how can you fault them for assuming you are going to use the RC. Furthermore, who are you to say that they can not discuss how to implement this for the RC, which the majority of the people reading this forum would benefit from. Yeah, its not completely centered on topic, but its still related to the original post, and the discussions here benefit a much larger group of people than just yourself.

Also, caps are considered "screaming" online, and it is pretty rude to use them.

Just my $.02.

[KenWittlief]

1. less then 200mA

2. as little noise as possible i will just be using normal TTL ICs

3. as efficient as possible, but effeciency for what? voltage?

4.as much room as as needed

5. through hole components

6. probaly a bread-board to test it then finalize it on a printed circuit board]

in this case the 7805 circuit already discussed is going to be the best approach.

efficiency measures how much power is lost or waisted. You normally only worry about efficiency if you are using a lot of power, or if you are running off batteries and really need to maximize battery life.

in this case, if you are dropping from 12V to 5, and drawing 200mA, then you are putting in 12*.2 = 2.4W and getting 1 watt out (5 * .2)

so this example is 1 / 2.4 => 42% efficient. Not great, but if your whole system is drawing 100 watts, or a 1000 watts (like a robot) then the lost 1.4W is not significant.

If you used a switching power supply instead you can get 90% efficiency or better, but the design is more complex, has more parts, and costs more.

[Gdeaver]

No specifics yet, so I describe a compressor setup we may use next year. Maybe everyone can critique it. We have a chip carrier board with the following.
1. A 7805 TO220 regulator non heat sinked.
2. A 10uf aluminum 16v cap on the input.
3. A 100uf aluminum cap on the output.
4. A MAX232A chip with bypass cap
5. A PIC 18F452 microcontroller with bypass cap.
6. 2 LEDs 20 ma each.
We have used the following senor chips - A analog devices gyro, last years accelerometer, a digital compass, a SRF04 sonar, a sharp IR ranger, and the hitec servo powered by the 5 volts. We have had up to 3 sensors at a time plus the servo. We have run it off of a 7.2 volt wall wart and a 7.2 volt 1800mah hobby nicad pack. Everything hardware wise is working fine. The set up has run for days and the regulator is just slightly warm. The total load max is < 350ma. Never ran it on 12 volts. So the question is what should be done for heat sinking? Is there something else that should be added to protect it from the robot battery? Spiky noise from the motors etc. What about fusing? What are the issues with the rs232 cable and the grounds. I a hobbyist not an EE so simple answers, don't assume to much. From what we have so far coprossesors look like a good way to extend the auto mode with out running into interrupt hell and other programing issues with the FRC.