I was wondering if there would be any real advantage to a linear PS design in using two regulators in series, say two LT108x types. The output of the first would be the input to the second. It seems like there might be some potential positive side effects (noise?), but apart from additional cost I suspect there are probably downsides as well. Any guru enlightenment appreciated.
I'm no guru, but:
Yes, it's a good idea, and is common enough that you can even find it in the LM317 datasheet on page 19.
http://cache.national.com/ds/LM/LM117.pdf
This technique is described as a "tracking preregulator" and you can search on this if you like. Here's a really good thread:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=28978
Yes, it's a good idea, and is common enough that you can even find it in the LM317 datasheet on page 19.
http://cache.national.com/ds/LM/LM117.pdf
This technique is described as a "tracking preregulator" and you can search on this if you like. Here's a really good thread:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=28978
sawreyrw,
Take a second look. There are some advantages... real ones. With two regs, one can handle line variations, the other load variations. If done right, with sufficient capacitance between the regs, significant gains can be made in line rejection.
Look at the graphs for line rejection versus frequency for typical 3 term regs. They get pretty lame around at 20 kHz; 40 dB rejection being common. You can effectively double this by using regs in series.
Now whether or not a given application requires this is another story altogether. An amp with 80dB PSRR across the band probably woulkd benefit little.
Take a second look. There are some advantages... real ones. With two regs, one can handle line variations, the other load variations. If done right, with sufficient capacitance between the regs, significant gains can be made in line rejection.
Look at the graphs for line rejection versus frequency for typical 3 term regs. They get pretty lame around at 20 kHz; 40 dB rejection being common. You can effectively double this by using regs in series.
Now whether or not a given application requires this is another story altogether. An amp with 80dB PSRR across the band probably woulkd benefit little.
poobah said:
Now that's a good idea I hadn't thought of. The requirements for line and load regulation are rather different, so it ought to be possible to better the doubling of rejection you've suggested, even if you design the first regulator to be a low drop-out voltage regulator (to minimise the inefficiency alluded to by sawreyrw).
Yep,
Along slighlty different lines... one of my favorite tricks for distributing power and signals from board to board in bundled wiring is to send out a constant current from the PSU. Then, placed a shunt reg. at the point of use. This keeps the power current in the bundled group stable and balanced.
It does present some problems with open sockets and hot-swap boards etc... you need to clamp your upstream current source.
I think JC uses the same concept in the Blowtorch, but for other reasons. The concept of splitting the line/load regulation in two remains constant though.
Do Anode and Cathode like the new venue?
Along slighlty different lines... one of my favorite tricks for distributing power and signals from board to board in bundled wiring is to send out a constant current from the PSU. Then, placed a shunt reg. at the point of use. This keeps the power current in the bundled group stable and balanced.
It does present some problems with open sockets and hot-swap boards etc... you need to clamp your upstream current source.
I think JC uses the same concept in the Blowtorch, but for other reasons. The concept of splitting the line/load regulation in two remains constant though.
Do Anode and Cathode like the new venue?
Hi poobah,
There is a time and place for everything. You can use different devices in series, each optimized for their job. That includes discrete units.
For voltage distribution (with local regulation) as you were saying, that's what they used to do in earlier computers. Each board had it's own regulation. I think there is merit in that approach.
Some phones systems have "hot swapable" boards. The only one I trust is the Eon system where you take the card off-line first. Local regulators.
EC8010,
Heater supplies could use a double regulation approach. One to get rid of HF "stuff" and the other to set the voltage. Same can be done for HV sections.
-Chris
There is a time and place for everything. You can use different devices in series, each optimized for their job. That includes discrete units.
For voltage distribution (with local regulation) as you were saying, that's what they used to do in earlier computers. Each board had it's own regulation. I think there is merit in that approach.
Some phones systems have "hot swapable" boards. The only one I trust is the Eon system where you take the card off-line first. Local regulators.
EC8010,
Heater supplies could use a double regulation approach. One to get rid of HF "stuff" and the other to set the voltage. Same can be done for HV sections.
-Chris
Poobah is quite correct. There can be value added for cascading voltage regulators. They are benificial for any desiqn that requires low noise/ripple in a noisy environment. Examples of their application can be found on PLL's, ADC, DAC, VCO's, and low noise high gain preamps on & on. The second stage Vreg is usually a discrete design in the most demanding applications. The bulk caps are not really able to cope with HF noise and ripple because of their higher esr and esl ratings.
two in series
" ... was wondering if there would be any real advantage to a linear PS design in using two regulators in series ..."
Yes, this can be useful:
30 VDC source regulated to 15 VDC and a second regulator for ~5 VDC to ~1.0 VDC.
Most 5 VDC fixed regulators have max input ratings of 24 volts, so in series with a 15 VDC regulator, say the plus side regulator of a dual rail +/- 15 VDC, ... appropriate balancing may be required ... but it works fine and the second tier regulation is very clean, low noise.
" ... was wondering if there would be any real advantage to a linear PS design in using two regulators in series ..."
Yes, this can be useful:
30 VDC source regulated to 15 VDC and a second regulator for ~5 VDC to ~1.0 VDC.
Most 5 VDC fixed regulators have max input ratings of 24 volts, so in series with a 15 VDC regulator, say the plus side regulator of a dual rail +/- 15 VDC, ... appropriate balancing may be required ... but it works fine and the second tier regulation is very clean, low noise.
BINGO... give those men a beer! Who said ESL & ESR?... 2 beers!
sawrey,
If were caps were caps... things would be so simple. Dude, I'm on your side... no silver wire, teflon boards, and pointy wooden things on the bottom of the CDP for me.
This one is worth a dig though... keep in mind, I did qualify this by saying that much depends on the PSRR of your load. The better that is, the less of problem regulation and noise blocking become. But alot of "zen" circuitry abounds these days, some with good reason. Many of these designs place higher demands on the PSU. Without a noise spec... no one can say what's wrong or right anyway.
I admire people putting $15 line filters on their gear... rather than $1500 power cords. The power line gets dirtier by the day. While we will be pushed into PFC power supplies that will preserve the sine... the hash goes up as well.
sawrey,
If were caps were caps... things would be so simple. Dude, I'm on your side... no silver wire, teflon boards, and pointy wooden things on the bottom of the CDP for me.
This one is worth a dig though... keep in mind, I did qualify this by saying that much depends on the PSRR of your load. The better that is, the less of problem regulation and noise blocking become. But alot of "zen" circuitry abounds these days, some with good reason. Many of these designs place higher demands on the PSU. Without a noise spec... no one can say what's wrong or right anyway.
I admire people putting $15 line filters on their gear... rather than $1500 power cords. The power line gets dirtier by the day. While we will be pushed into PFC power supplies that will preserve the sine... the hash goes up as well.
I don't want this to get silly, but most people know you put good high quality RF (.1uf, etc.) caps across all loads that need them. This is in addition to the bulk caps across the regulator.
I know it depends on the PSRR of the load; just show me a real example where 2 series regulators in series make sense. Oh, I know about line frequency 'preregulators' used a couple of decades ago.
I know it depends on the PSRR of the load; just show me a real example where 2 series regulators in series make sense. Oh, I know about line frequency 'preregulators' used a couple of decades ago.
$15 v. $1500 line filters and gear
Yes indeede . ... spending big bucks of line filters can be a serious waste of resources ...
In the case of line filters, US$15 is about all any DIYer should consider ... I made mine from a simple circuit picked up from an article in ARRL Handbook circa 1965 ... hand made air core coil of #14 AWG, wrapped around a part of a broom stick, insulated with real good tape plus a 600 volt capacitor = LC time constant in the 2k Hertz range = shunts everything not useful as pure 60 to 120 Htz. power to the white wire & ground, inside a 4X utility box with industrial / hospital duplex outlets.
Yes indeede . ... spending big bucks of line filters can be a serious waste of resources ...
In the case of line filters, US$15 is about all any DIYer should consider ... I made mine from a simple circuit picked up from an article in ARRL Handbook circa 1965 ... hand made air core coil of #14 AWG, wrapped around a part of a broom stick, insulated with real good tape plus a 600 volt capacitor = LC time constant in the 2k Hertz range = shunts everything not useful as pure 60 to 120 Htz. power to the white wire & ground, inside a 4X utility box with industrial / hospital duplex outlets.
makes sense to me ...
" ... just show me a real example where 2 series regulators in series make sense. ..."
As above ...
Consider a well regulated system with +/- 15 VDC for the amplifier rails ... suppose you wanted to add a digital VU meter and the circuit called for well regulated reference voltage. Rather than use a reference Zener (wasteful of power), you could substitute a fixed (or adjustable) linear regulator as voltage reference for the meters, driven from the +15 rail ...
Another:
Consider the industry standard PC ... has switching power supplies with large amounts of regulated 5 VDC .... these feed the motherboard and somewhere around the main CPU processor (or right onboard the chip) is a second regulator (usually a DC to DC converter, but a genuine regulator none the less) that knocks the voltage down to 3.3 VDC ...
" ... just show me a real example where 2 series regulators in series make sense. ..."
As above ...
Consider a well regulated system with +/- 15 VDC for the amplifier rails ... suppose you wanted to add a digital VU meter and the circuit called for well regulated reference voltage. Rather than use a reference Zener (wasteful of power), you could substitute a fixed (or adjustable) linear regulator as voltage reference for the meters, driven from the +15 rail ...
Another:
Consider the industry standard PC ... has switching power supplies with large amounts of regulated 5 VDC .... these feed the motherboard and somewhere around the main CPU processor (or right onboard the chip) is a second regulator (usually a DC to DC converter, but a genuine regulator none the less) that knocks the voltage down to 3.3 VDC ...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.