I am attaching a linear design I intend to put together. I wanted to get input from the gurus out here, in particular with regard to the sizes and types of caps. Even though I don't have the bridge rectifier marked, it will actually be made up of four Fairchild TO220 "soft recovery" type diodes, with appropriate heat sinks. I don't have the transformer yet, but I'm looking at varieties in the 10 VAC secondary, 3.8 amp out category. Input is standard 120VAC, 60Hz. The output of the circuit is meant to be 5VDC, capable of driving up to 3A. In reality it will be supplying about 1.2A on average but should be able to peak up to just under 3.
The other thing to note going in is that I intend to build this so that the transformer, the PCB containing the bridge + C1 + C2, and the remainder of the circuit will be physically separated from each other but in one enclosure. I will put shielding around each section to reduce electrical and magnetic junk from either the tranny or the regulator diodes from getting into each section 2. So there will be leads from the tranny to section 1, and from section 1 to section 2 (between C2 and C3).
Here's what I've thought about using for the various capacitors:
C1 - either a Jensen 4 pole 15000uf, 25V or (4) 4700uF Panasonic FC types
C2 - a 10uF bypass, but I don't know if I should use something like an Auricap polyprop, or a tantalum. And should I use another even smaller value (1 or .1) on top of the 10?
C3 - either the same type Jensen 4 pole 15000uF, or their 4 pole 47000 uF. I'm doing this because, again, this section will be separated from the other, but I'm not sure if I need another bypass here.
C4, C5 - LT recommends on the order of 20-25uF for decent ripple rejection at 120 Hz, but then goes on to say that a .22uF would provide rejection at the higher frequency range (10KHz). I see I also made a mistake here. These caps should be electrolytic, but here's where I could use the greatest help with both value(s) and type.
C6 - Jensen 4 pole, 47000uF
C7 - Sonicap or Auricap bypass, but I'm not sure if I need more than one (10uF + .1uF) or just go with 1 (1uF).
R1 and R2 will be selected to generate the required output voltage. Probably use Dale low noise types or Caddocks.
It should be obvious to those with greater knowledge that I have just enough understanding to get myself in trouble.
So I turn to you all to help me get the best I can out of this basic design. Thanks in advance.
The other thing to note going in is that I intend to build this so that the transformer, the PCB containing the bridge + C1 + C2, and the remainder of the circuit will be physically separated from each other but in one enclosure. I will put shielding around each section to reduce electrical and magnetic junk from either the tranny or the regulator diodes from getting into each section 2. So there will be leads from the tranny to section 1, and from section 1 to section 2 (between C2 and C3).
Here's what I've thought about using for the various capacitors:
C1 - either a Jensen 4 pole 15000uf, 25V or (4) 4700uF Panasonic FC types
C2 - a 10uF bypass, but I don't know if I should use something like an Auricap polyprop, or a tantalum. And should I use another even smaller value (1 or .1) on top of the 10?
C3 - either the same type Jensen 4 pole 15000uF, or their 4 pole 47000 uF. I'm doing this because, again, this section will be separated from the other, but I'm not sure if I need another bypass here.
C4, C5 - LT recommends on the order of 20-25uF for decent ripple rejection at 120 Hz, but then goes on to say that a .22uF would provide rejection at the higher frequency range (10KHz). I see I also made a mistake here. These caps should be electrolytic, but here's where I could use the greatest help with both value(s) and type.
C6 - Jensen 4 pole, 47000uF
C7 - Sonicap or Auricap bypass, but I'm not sure if I need more than one (10uF + .1uF) or just go with 1 (1uF).
R1 and R2 will be selected to generate the required output voltage. Probably use Dale low noise types or Caddocks.
It should be obvious to those with greater knowledge that I have just enough understanding to get myself in trouble.
So I turn to you all to help me get the best I can out of this basic design. Thanks in advance.Attachments
C7 & C2 should be TANTALUM caps
C7 should be around 22uF
C2 should be around 10uF
if you want to keep the supply quiter put a .1 ceramic cap on the AC inputs of your bridge rectifier, maybe even one on the reverse protection diode on your regulator?
I not really sure you need such a big cap (or Caps) on the output of the reg since the main filter cap will get ride of any ripple.
C7 should be around 22uF
C2 should be around 10uF
if you want to keep the supply quiter put a .1 ceramic cap on the AC inputs of your bridge rectifier, maybe even one on the reverse protection diode on your regulator?
I not really sure you need such a big cap (or Caps) on the output of the reg since the main filter cap will get ride of any ripple.
the LT1085 is a low dropout regulator so caps C6 and C7 are in the control loop -- Linear specifies a 10uF tantalum -- you can probably parallel an electrolytic and tantalum, but a high value electrolytic on its own (as IFRYTHINGS points out) should be avoided as it can cause loop instability -- i.e. you can get the regulator to oscillate.
Putting caps across diodes has a small effect upon noise -- these are in parallel with the diode's junction capacitance so any ringing in the tank circuit is lowered in frequency and the voltage spike across the diode is somewhat smaller -- I prefer to use an RC network across the diode which knocks out the ringing of the tank circuit. Whether you have a ringing in the diode circuit -- you will need a good oscilliscope to see it -- depends upon the leakage inductance of the transformer you are using, interwinding capacitance etc. In some applications these are completley un-necessary.
If anything, the best thing you can do is plan the layout carefully -- Linear has a design using this family (see page 12) with a separate error amplifier -- this would probably be much more effective in implementation and far less expensive than using a filter cap which costs tens of dollars.
Putting caps across diodes has a small effect upon noise -- these are in parallel with the diode's junction capacitance so any ringing in the tank circuit is lowered in frequency and the voltage spike across the diode is somewhat smaller -- I prefer to use an RC network across the diode which knocks out the ringing of the tank circuit. Whether you have a ringing in the diode circuit -- you will need a good oscilliscope to see it -- depends upon the leakage inductance of the transformer you are using, interwinding capacitance etc. In some applications these are completley un-necessary.
If anything, the best thing you can do is plan the layout carefully -- Linear has a design using this family (see page 12) with a separate error amplifier -- this would probably be much more effective in implementation and far less expensive than using a filter cap which costs tens of dollars.
Jack, I'm taking all this good advice in. I tended toward large caps (suitably bypassed, of course) mostly because I'm trying to duplicate a "high end" aftermarket supply. The company that offers this supply also uses a LT1085 and has indicated that in their long experimentation with different cap values they settled on the 47000 uF jobs. Did they use SPICE during the design phase and quantitative testing afterwards? I doubt it, but people I know who are using this PS all indicate that it's pretty good.
Are you referring to the "Remote Sensing" layout on page 12 here?
http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1040,C1055,P1283,D3741
Are you referring to the "Remote Sensing" layout on page 12 here?
http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1040,C1055,P1283,D3741
the problem with a big electrolytic in an application like this is that the voltage will hang high, or it will take a big charge to get it up to the desired level -- you can work out the math -- but a high value overdamps the circuit.tonyptony said:
the interest I had in the remote sense circuit is that the error control is directly associated with the application -- note the Sense + and Sense - inputs --
but when you start adding to the complexity you might as well go down the "discrete route" and build a Jung regulator with a fast, low noise error amplifier, etc.
jackinnj said:
Jack, that's part of the reason I'm taking a somewhat simpler approach. I agree about the overdamping that is introduced with big caps, but I'm taking the brute force route - hoping that the right selection of components down this road will get me the best solution this sort of design can bring.
Tantalums
C2 should NOT be tantalum. I consider this a fault in the datasheet. In this position it has high ripple on it and it does not like it. One burned itself to a glowing coal!
In the past I used a lot of tantalums but I am phasing them out in all my designs. They often fail with a lot of smoke and bad smell. I like Oscons when low ESR is needed.
C2 should NOT be tantalum. I consider this a fault in the datasheet. In this position it has high ripple on it and it does not like it. One burned itself to a glowing coal!
In the past I used a lot of tantalums but I am phasing them out in all my designs. They often fail with a lot of smoke and bad smell. I like Oscons when low ESR is needed.
Re: Tantalums
I couldn't get the LT1763 to work with anything BUT a tantalum -- everything else made it oscillate -- of course we are talking TL1083,4,5 here.
Elso Kwak said:C2 should NOT be tantalum. I consider this a fault in the datasheet. In this position it has high ripple on it and it does not like it. One burned itself to a glowing coal!
In the past I used a lot of tantalums but I am phasing them out in all my designs. They often fail with a lot of smoke and bad smell. I like Oscons when low ESR is needed.
I couldn't get the LT1763 to work with anything BUT a tantalum -- everything else made it oscillate -- of course we are talking TL1083,4,5 here.
Hi Jackinnj,
C1 & C2 are parallel.
Are you saying that the lower esr of the tantalum will take more than it's share of ripple because the electro has a higher esr?
I would have thought the ripple would be proportioned by ratio of their capacitance, 15,000uf to 10uF.
Is ripple in the tantalum in this // location critical?
C1 & C2 are parallel.
Are you saying that the lower esr of the tantalum will take more than it's share of ripple because the electro has a higher esr?
I would have thought the ripple would be proportioned by ratio of their capacitance, 15,000uf to 10uF.
Is ripple in the tantalum in this // location critical?
Re: Tantalums
It's fairly rare to find tantalums used in domestic electronics, they were common for a few years but they proved EXTREMELY unreliable - as suggested, the common failure mode is S/C.
Standard fault finding practice - if you see a tantalum, suspect it's S/C before you test anything else!.
Elso Kwak said:
It's fairly rare to find tantalums used in domestic electronics, they were common for a few years but they proved EXTREMELY unreliable - as suggested, the common failure mode is S/C.
Standard fault finding practice - if you see a tantalum, suspect it's S/C before you test anything else!.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.