what is the best positive (3 pin only ) 5v voltage regulator? i've read about the lt1085, but the website says it is an old model and do not recomend it (??).. are there any better/newer options.
the ones in place now are (3) 78L05's each bypassed by one 47uf electrolytic before and one 47uf plus a 47nf after the regulator...should i replace this caps values? i was planning using oscon the same value ...
the regulator is for the dac chip in a cd player.
m.
the ones in place now are (3) 78L05's each bypassed by one 47uf electrolytic before and one 47uf plus a 47nf after the regulator...should i replace this caps values? i was planning using oscon the same value ...
the regulator is for the dac chip in a cd player.
m.
Not to disagree with anyone but I have had excellent results with the 78xx series for different applications. The design criteria that I use is a .01uf ceramic cap at the input and 6 .01uf ceramic caps followed by a 10uf tantulum cap on the output...this provides a very low impedance and the noise is reduced quite alot.
Cheers!!The DIRT®
Cheers!!The DIRT®
Three terminal only? Discrete regulators are a lot of fun, and you can build them with three pins on one side if you want
---
there is very good info about discrete designs (i already read some) but and i am looking for something easy to implement (im a new with the iron you know
) that is the reason why i am looking for direct replacement (fixed voltage 3 pin) ...there has to be a good 3 pin, 5v regulator for digital aplications somewhere ...eventually i will build something more complicated..
---
I like this one.
i will order some ..thanks
---
there is very good info about discrete designs (i already read some) but and i am looking for something easy to implement (im a new with the iron you know
) that is the reason why i am looking for direct replacement (fixed voltage 3 pin) ...there has to be a good 3 pin, 5v regulator for digital aplications somewhere ...eventually i will build something more complicated..---
I like this one.
i will order some ..thanks
Re: Re: Re: +5v regulator
I have not found that swapping 3 pin regulators made much difference in the sound, presumably because they don't perform much better than those I have measured, and because they don't have low enough impedance broadband. What I am a bit cut up about (having followed advice for sometime) is that the LT series is better. They cost a lot more and are a pain because of the non standard pin arrangement.
If you want best souind, you need to use discrete regulators, perhaps pay Audiocom £40 for a regulator, or design your own.
-------------------------------------------------Peter Daniel said:
But how did they sound? Measurements mean nothing.
I'm asking, because I'm also looking for a good regulator and don't want to go through the hassle of listening to all of them
I have not found that swapping 3 pin regulators made much difference in the sound, presumably because they don't perform much better than those I have measured, and because they don't have low enough impedance broadband. What I am a bit cut up about (having followed advice for sometime) is that the LT series is better. They cost a lot more and are a pain because of the non standard pin arrangement.
If you want best souind, you need to use discrete regulators, perhaps pay Audiocom £40 for a regulator, or design your own.
Koinichiwa,
I found the LT/LM1085-5 (the fixed Voltage Version of the 1085) to be NO improvement on normal 3-pin regs. There is a simple reason too. The build-in resistor divider has not got the option to bypass the ADJ-2-GND resistor, which in such regulators actually drastically lowers noise and output impedance.
Now you could of course use the Adjustable version of the 1085 and incude the two needed resistors and the bypass cap, in which case it would actually be better than any 7805 I have come across, but ONLY then.
My personal preference for a "precision" 3-Terminal regulator circuit is attached. This is suited of course only to relatively low currents, with a suitable CCS (selected J-Fet) it can be used up to past the 50mA output current region.
If you don't mind exceeding the LM334 rated max current (I have done it a few times with no ill effects) you can have more output current even using the LM334, simply stay well within the dissipation rating of the LM334.
The actual regulator is the bit in the Box, the three decoupling Cap's are my recommended local decoupling scheme regardless of regulator used (7805 or the like do need a larger value Os-Con, more like 100uF++ though).
The actual advantage over the usual 3-pin types is usually realised in noise AND much more imporatntly in circuit if the regulators are local to a given chip the Impedance, which remains constant and low up to around 100KHz.
The Os-Con extends this wide range of low PSU impedance to around 500...600KHz. The other SMD caps IF APPLIED WITH MINIMAL LOOP AREAS keep the impedance low past the 30MHz mark...
I find that wiring up this circuit using the two TO-92 cases superglued together and using 1206 SMD resistors makes for a VERY compact circuit that can be fitted instead of most 3-Pin regulators and is not THAT difficult to make.
I also always have ferrite beads on both input to and output from the CCS, this helps to kill feedthrough at higher frequencies. Adding a series inductor somewhere on the board before the trace joins into the main PSU bus is also a good idea. Often the Designer has already thoughtfully (NOT) provided suitable Wirebridges that can be replaced by radial or axial inductors. If not you can often fit an SMD inductor into a PCB Trace.
Of course, this circuit can be bettered but with it's 5-part simplicity, constant current draw and short AC loops it offers much better insulation of PSU lines against each other and noise and that with components that are cheap and easily obtained. It is not a universal Reg but simple and often applicable in modding existing gear.
Anyway, my experiences and thoughs only.
Sayonara
mekanoplastik said:
I found the LT/LM1085-5 (the fixed Voltage Version of the 1085) to be NO improvement on normal 3-pin regs. There is a simple reason too. The build-in resistor divider has not got the option to bypass the ADJ-2-GND resistor, which in such regulators actually drastically lowers noise and output impedance.
Now you could of course use the Adjustable version of the 1085 and incude the two needed resistors and the bypass cap, in which case it would actually be better than any 7805 I have come across, but ONLY then.
My personal preference for a "precision" 3-Terminal regulator circuit is attached. This is suited of course only to relatively low currents, with a suitable CCS (selected J-Fet) it can be used up to past the 50mA output current region.
If you don't mind exceeding the LM334 rated max current (I have done it a few times with no ill effects) you can have more output current even using the LM334, simply stay well within the dissipation rating of the LM334.
The actual regulator is the bit in the Box, the three decoupling Cap's are my recommended local decoupling scheme regardless of regulator used (7805 or the like do need a larger value Os-Con, more like 100uF++ though).
The actual advantage over the usual 3-pin types is usually realised in noise AND much more imporatntly in circuit if the regulators are local to a given chip the Impedance, which remains constant and low up to around 100KHz.
The Os-Con extends this wide range of low PSU impedance to around 500...600KHz. The other SMD caps IF APPLIED WITH MINIMAL LOOP AREAS keep the impedance low past the 30MHz mark...
I find that wiring up this circuit using the two TO-92 cases superglued together and using 1206 SMD resistors makes for a VERY compact circuit that can be fitted instead of most 3-Pin regulators and is not THAT difficult to make.
I also always have ferrite beads on both input to and output from the CCS, this helps to kill feedthrough at higher frequencies. Adding a series inductor somewhere on the board before the trace joins into the main PSU bus is also a good idea. Often the Designer has already thoughtfully (NOT) provided suitable Wirebridges that can be replaced by radial or axial inductors. If not you can often fit an SMD inductor into a PCB Trace.
Of course, this circuit can be bettered but with it's 5-part simplicity, constant current draw and short AC loops it offers much better insulation of PSU lines against each other and noise and that with components that are cheap and easily obtained. It is not a universal Reg but simple and often applicable in modding existing gear.
Anyway, my experiences and thoughs only.
Sayonara
Attachments
Re: Re: +5v regulator
My personal preference for a "precision" 3-Terminal regulator circuit is attached. This is suited of course only to relatively low currents, with a suitable CCS (selected J-Fet) it can be used up to past the 50mA output current region.
---------------------------------------------------
7 mA in your circuit is too little for most applications. Can you post your add to push current to 50 or even 100 mA. 300 mA will be even more useful
My personal preference for a "precision" 3-Terminal regulator circuit is attached. This is suited of course only to relatively low currents, with a suitable CCS (selected J-Fet) it can be used up to past the 50mA output current region.
---------------------------------------------------
7 mA in your circuit is too little for most applications. Can you post your add to push current to 50 or even 100 mA. 300 mA will be even more useful
Re: Re: Re: +5v regulator
Koinichiwa,
The LM334 is rated at nominally at 10mA max. current. HOWEVER, I have used the LM334 with no ill effects with as much as 30mA and would not see any significant problems with higher currents even, as long as the dissipation is being kept below around 200mW (without added cooling fins). This means with 10V across the LM334 you are limited to 20mA.
The resistor to set the current through the LM334 is approximatly 68R/I (in ma). As you can see I have set the current to nearly 14mA, following the classic formula of I(ccs) = 2* I(load).
For higher current I think simply choosing a J-Fet with Idss = Iout *1.5 will be a good option. J-Fets with suitable ratings exist. Alternatively an LM317 or LM1085 can be used in current souce configuration, the current set resistor then becomes 1250/I (in mA).
However, most modern digital chips have very low real current consumption anyway. You can usually get away with LM334 & TL/LM431. If there is much more supply current I normally suggest a totally different approach whcih is however not really suited to inexperienced DIY'er or those without extensive measurement gear (possible instability).
In the circuit discussed above at the beginning of the thread the total current drawn by the PCM1716 is only 45mA max (at 96KHz) for all three supply nodes together. No specific values are given, but I'd suspect that each node draws no more than 30mA where the 30mA are the digital section under worst case conditions.
So arranging each CCS for 40mA (R for LM334 = 1R8, R for LM317/1085 = 33R) or selecting a J-Fet with Idss = 40mA will be fine. In case the LM334 is used as CCS a small add-on/clip-on heatsink would be strongly suggested.
Sayonara
Koinichiwa,
fmak said:
The LM334 is rated at nominally at 10mA max. current. HOWEVER, I have used the LM334 with no ill effects with as much as 30mA and would not see any significant problems with higher currents even, as long as the dissipation is being kept below around 200mW (without added cooling fins). This means with 10V across the LM334 you are limited to 20mA.
The resistor to set the current through the LM334 is approximatly 68R/I (in ma). As you can see I have set the current to nearly 14mA, following the classic formula of I(ccs) = 2* I(load).
For higher current I think simply choosing a J-Fet with Idss = Iout *1.5 will be a good option. J-Fets with suitable ratings exist. Alternatively an LM317 or LM1085 can be used in current souce configuration, the current set resistor then becomes 1250/I (in mA).
However, most modern digital chips have very low real current consumption anyway. You can usually get away with LM334 & TL/LM431. If there is much more supply current I normally suggest a totally different approach whcih is however not really suited to inexperienced DIY'er or those without extensive measurement gear (possible instability).
In the circuit discussed above at the beginning of the thread the total current drawn by the PCM1716 is only 45mA max (at 96KHz) for all three supply nodes together. No specific values are given, but I'd suspect that each node draws no more than 30mA where the 30mA are the digital section under worst case conditions.
So arranging each CCS for 40mA (R for LM334 = 1R8, R for LM317/1085 = 33R) or selecting a J-Fet with Idss = 40mA will be fine. In case the LM334 is used as CCS a small add-on/clip-on heatsink would be strongly suggested.
Sayonara
I've been using a similar arrangement like Thorsten, using jfets for the CCS's. I also implemented a LM317/LM413 combo for higher currents (don't disconnect load or LM413 will blow).
I've used J309's for currents up to 20mA and recently J111's with resistor between source and gate to set the current.
For high output impedances, choose Vds as high as possible, not exceeding the thermal dissipation limits of the fet though.
For even "better" ccs's one can cascade combinations of fets, LM334, LM317's, bjt etc.
I've used J309's for currents up to 20mA and recently J111's with resistor between source and gate to set the current.
For high output impedances, choose Vds as high as possible, not exceeding the thermal dissipation limits of the fet though.
For even "better" ccs's one can cascade combinations of fets, LM334, LM317's, bjt etc.
Looped beads
Kuei,
On post #15 you mentioned the use of ferrite beads at the input and output of the local regulator.
As far as my knowledge goes, the beads will limit the RF through the wire and this is atained because the wire through the bead forms a small inductor. OK so far?
Well if this is true, using a somewhat bigger inductor would provide better isolation I suppose.
I mesured the inductance provided by a single bead used in the regular way and got 0.7 uH. Now if we make a single loop over the ferrite we get something like 3.1 uH.
Would the use of the looped ferrite bead provide better RF isolation? Speaking of low load as applied to DACs, say 25/30 mA
Kuei,
On post #15 you mentioned the use of ferrite beads at the input and output of the local regulator.
As far as my knowledge goes, the beads will limit the RF through the wire and this is atained because the wire through the bead forms a small inductor. OK so far?
Well if this is true, using a somewhat bigger inductor would provide better isolation I suppose.
I mesured the inductance provided by a single bead used in the regular way and got 0.7 uH. Now if we make a single loop over the ferrite we get something like 3.1 uH.
Would the use of the looped ferrite bead provide better RF isolation? Speaking of low load as applied to DACs, say 25/30 mA
Re: Looped beads
Koinichiwa,
Yes, a few loops are likely better, even better is a real inductor (a few mH) in series ahead of the regulator (also helps for 3-Pin regs BTW). You can also further increase the isolation by cascoding the current-source.
Once you are willing to increase complexity you can do a lot of things, but then there are also other options. My original aim was to simply "KISS" (Keep It Simple St...d).
Sayonara
Koinichiwa,
apassgear said:
Yes, a few loops are likely better, even better is a real inductor (a few mH) in series ahead of the regulator (also helps for 3-Pin regs BTW). You can also further increase the isolation by cascoding the current-source.
Once you are willing to increase complexity you can do a lot of things, but then there are also other options. My original aim was to simply "KISS" (Keep It Simple St...d).
Sayonara
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.