I have been trying to figure out for a all around power supply is using a LM317 at Vo = 5V better then using a 7805. I have compiled a spreadsheet with data from 19 different ICs and it appears a 7805 is a better solution. I know this is discussed here a lot and thought you guys might like such a collection of information.
Does anyone have any real world experience with this subject or opinions in general?
Thanks,
Matt
P.S. A related post of mine with more information can be found at http://groups.yahoo.com/group/techlibdotcom/message/505
Does anyone have any real world experience with this subject or opinions in general?
Thanks,
Matt
P.S. A related post of mine with more information can be found at http://groups.yahoo.com/group/techlibdotcom/message/505
Hi Matt,
The data sheets are correct. Many experimenters have joined "the bigger capacitor is better" camp. This without doing any real studies beyond possibly a listening test.
One assumes that you will be supplying the regulator with filtered DC. Therefore, the capacitor immediately before the regulator is there to provide a low AC impedance at higher frequencies. Assuming that the regulator is within a few inches of your filters you are okay. If you have greater distance (read higher impedance) between the filters and the regulator, you should add a larger electrolytic close to the regulator as well. This occurs when you have local regulation on a card in a larger system.
The capacitor at the output does two things. It provides a low AC impedance for the load at higher frequencies and it supplies current if the load draws current with sudden surges. The output impedance of a regulator increases with frequency like an amplifier. Well, it does contain an amplifier so this should not be a surprise. Also, there is a fair amount of gain. It may oscillate if there is no bypassing close by (your small capacitors). Larger capacitors have inductance and do not generally work well at the higher frequencies we are concerned with. By omitting the bypass caps, they will probably have an oscillating regulator at some point in time. Electrolytic capacitors normally age with increasing internal resistance and a drop in capacitance.
Semiconductor companies spend a great deal of money and time designing and researching how to use these products. They will be right more often than wrong. They also seem to agree, so that means their engineers agree also. I'm betting they are dead on the money here.
Now, it is possible that a defect in PCB layout or circuit design may require larger capacitance on the output of a regulator. That does not mean that the smaller capacitors can be removed. It does mean that you run the danger of having your supply discharge backwards through the regulator when the power is turned off. This can and does happen. A reverse biased diode across the regulator is the protection from the that occurring.
-Chris
The data sheets are correct. Many experimenters have joined "the bigger capacitor is better" camp. This without doing any real studies beyond possibly a listening test.
One assumes that you will be supplying the regulator with filtered DC. Therefore, the capacitor immediately before the regulator is there to provide a low AC impedance at higher frequencies. Assuming that the regulator is within a few inches of your filters you are okay. If you have greater distance (read higher impedance) between the filters and the regulator, you should add a larger electrolytic close to the regulator as well. This occurs when you have local regulation on a card in a larger system.
The capacitor at the output does two things. It provides a low AC impedance for the load at higher frequencies and it supplies current if the load draws current with sudden surges. The output impedance of a regulator increases with frequency like an amplifier. Well, it does contain an amplifier so this should not be a surprise. Also, there is a fair amount of gain. It may oscillate if there is no bypassing close by (your small capacitors). Larger capacitors have inductance and do not generally work well at the higher frequencies we are concerned with. By omitting the bypass caps, they will probably have an oscillating regulator at some point in time. Electrolytic capacitors normally age with increasing internal resistance and a drop in capacitance.
Semiconductor companies spend a great deal of money and time designing and researching how to use these products. They will be right more often than wrong. They also seem to agree, so that means their engineers agree also. I'm betting they are dead on the money here.
Now, it is possible that a defect in PCB layout or circuit design may require larger capacitance on the output of a regulator. That does not mean that the smaller capacitors can be removed. It does mean that you run the danger of having your supply discharge backwards through the regulator when the power is turned off. This can and does happen. A reverse biased diode across the regulator is the protection from the that occurring.
-Chris
In your sheet, why are there duplicate sets of specs for the same device? The columns I pasted below show the ripple rejection is either 68.0 or 83.0.
ON MC7805B ON MC7805C
35 35
2.2 2.2
1.3 0.8
0.15 1.3
N/A N/A
3.2 3.2
50 50
68.0 83.0
2.0 2.0
125 125
65 65
-0.3 -0.3
ON MC7805B ON MC7805C
35 35
2.2 2.2
1.3 0.8
0.15 1.3
N/A N/A
3.2 3.2
50 50
68.0 83.0
2.0 2.0
125 125
65 65
-0.3 -0.3
Sorry, It looks like I didn't make the column large enough to show the last letter. On the sheet there is a MC7805A, MC7805B, and MC7805C.
I have decided for the project I am doing that an adjustable regulator is the way to go. Over the next few days I am going to try to add to the adjustable part of the spreadsheet. My search criteria will be as follows.
1A - 2.5A output
Vo at least 5V - 15V
TO-220 package
Any ICs you guys want to see on the spreadsheet?
I have decided for the project I am doing that an adjustable regulator is the way to go. Over the next few days I am going to try to add to the adjustable part of the spreadsheet. My search criteria will be as follows.
1A - 2.5A output
Vo at least 5V - 15V
TO-220 package
Any ICs you guys want to see on the spreadsheet?
Ah, right, I didn't see the A/B/C on the spreadsheet but it sure showed up when I pasted it. Looks like ON has improved the old 7805.
I used the yellow highlighting to differentiate the better values. That way I could a get a quick visual of what IC is the overall best because no one IC was the best at everything.
OK I thought it might have been that. Was just wondering as rows 19, 22 and 23 had a couple of different values highlighted.
richie00boy said:
Gordy said:
Your welcome.
I am no focusing my researching efforts towards adjustable regulators. I have compiled more data into the adjustable regulator section. The spreadsheet now has specs for 34 regulators. 23 adjustable and 11 fixed 7805. One adjustable that sticks out is the Linear LT3080. It has the following specs.
Max Input Voltage - 40V
Max Current Output - 1.4A
Line Regulation - 0.003mV/V
Load Regulation - 0.6mV
Noise - 40uV
Ripple Rejection - 75dB
Max Junction Temperature - 125C
Thermal Resistance to Air - 40C/W
Does anyone have any experience with the LT3080?
Or recommendations on other adjustable linear regulators?
I have updated the color scheme. Yellow is for better than average values and blue is the best in category.
Thanks,
Matt
After doing a cost analysis, I have concluded three linear adjustable regulators have great value.
Non critical areas - ST LM317 $0.61
High current needs - National LM350A $2.55
Critical areas - Linear LT3080 - $3.14
I have updated the spreadsheet to include the following six tabs for anyone interested.
* All Adjustable vs All 7805
* Best of the Best Adjustable
* Top Value
* Priority List
* Load Reg Conversion Table
* Line Reg Conversion Table
Non critical areas - ST LM317 $0.61
High current needs - National LM350A $2.55
Critical areas - Linear LT3080 - $3.14
I have updated the spreadsheet to include the following six tabs for anyone interested.
* All Adjustable vs All 7805
* Best of the Best Adjustable
* Top Value
* Priority List
* Load Reg Conversion Table
* Line Reg Conversion Table
What do you think of adding one transistor to a regulator, like this to alter ripple rejection:
PCB for teddy's 'Regulator booster' - pink fish media
120Hz ripple rejection is only one feature, and not so thrilling if that's all it does.
There are also much higher frequencies, and I did not see that information on your spreadsheet.
PCB for teddy's 'Regulator booster' - pink fish media
120Hz ripple rejection is only one feature, and not so thrilling if that's all it does.
There are also much higher frequencies, and I did not see that information on your spreadsheet.
Last edited:
For a low noise supply the venerable 723 is worth considering. The LM723 claims 2.5 uV typical measured using a bandwidth of 100 Hz to 10 kHz. Unfortunately not specified from 10 Hz to 100 kHz for a direct comparison. Ripple rejection 86 dB typical.
It's more difficult to use than the 3 terminal types but is scalable uisng external transistors to many amps.
It's more difficult to use than the 3 terminal types but is scalable uisng external transistors to many amps.
TheDriver, very good post! Thank you for doing that comparison, you saved me hours of research! 🙂
The datasheet gives you the RR graph for each chip. The RR for the MC7805C at 1kHz = 75dB, 10kHz =70db
The RR graph oscillates and it only goes up to 10kHz
Also OP is only comparing specs and not RL tests, but still a good baseline 🙂
A simple RC filter on the regulator input, dropping a volt or so, can really improve higher frequency ripple rejection - far more than using an expensive chip
+1 - no matter how expensive the chip, the specs get progressively more terrible the higher the frequency. Can't beat passive filtering (R or even L) on the input.
If the expensive chip isn't working out, why not use a cheap chip that does? >>>> OnSemi MC
For adding the inductive filter, howabout a gyrator like this: Teddy's 'Regulator booster' With perhaps NXP BCP53, BCP56 or similar if you needed some more strength?
I think that Teddy circuit is for sibilant sources since I have tried it and it seems to be overkill. What I actually used was CDC filter (somewhat like CRC filter), and that was pretty neutral and appreciable.
For adding the inductive filter, howabout a gyrator like this: Teddy's 'Regulator booster' With perhaps NXP BCP53, BCP56 or similar if you needed some more strength?
I think that Teddy circuit is for sibilant sources since I have tried it and it seems to be overkill. What I actually used was CDC filter (somewhat like CRC filter), and that was pretty neutral and appreciable.
Last edited:
- Status
- Not open for further replies.