Alcaid said:
1. The max current depends the power dissipated in the pass transistor, which depends on your input voltage (how much drop across the pass transistor), and your load current. You need to know both parameters to design the regulator. If the drop is say 10v (from your input voltage of 25v), and the current is 25ma, you have 250mw dissipation, pretty high for this little device. It is rated 500 mw, but only if you keep it cool. Better to use a power darlington, it can handle more power without heatsinking.
2. Use a 1W 1N4745A, 16 volt nominal zener.
3. yes!
4. It's fine, but you do not need R3 and R13. They ruin line (hum) regulation. Select a zener diode that has the proper voltage and use it alone. If you want a precise voltage output, you need to use a better design which has an adjustment pot, either with an op amp and reference like a TL431. Better yet -- use an LM317/LM337 pair. These can easily be trimmed, and are short circuit proof.
Alcaid said:
1 No, do not use diodes. The zener is intended for this function, simple regulators. Delete the resistors entirely.
2 Darlingtons like TI141/142 would be much better, more gain, more power.
3 There is no benefit to a dual mono.
4. The capacitance is not critical, if your regulator has sufficient ripple rejection for whatever it powers. Anything from 220-2200uF is probably fine, depending on load current and input voltage. If your input voltage is high, you can get away with less, but if it's only 20v, use more. A darlington will reduce ripple, and bigger filters in parallel with the zener will also help
Alcaid said:
First mod: assume for this discussion that the zeners are a voltage source with zero internal resistance. Then for AC, noise, hum etc the zener series resistance and the R3/R13 form a voltage divider, right? This divider puts a divided down part of the noise, hum, etc on the supply input on the base of the follower transistor, which faithfully reproduces this junk at the supply output. You don't want that, I guess. So, replacing R3/R13 with a diode, which, by and large, can also be considered as a voltage source with zero impedance, makes the voltage at the base independent of the junk at the supply input. That is what you want, isn't it?
Second mod: it's a matter of priority. Larger standing current gives better and more stable supply, but requires a bigger transistor. If using the BC's is more important than the supply quality, that's your choice.
Jan Didden
janneman said:
You mean BD139/BD140, right? 🙂
They can deliever more current than the BC550C/BC560C, which is good, but they do have lower beta. Is the beta high enough? (40-250 vs. 420-800)
Ok, janneman here goes:
Changes to current schematic:
1. Removed the resistors earlier referred to as R13.
2. Changed transistors to the BD139/BD140.
3. Added input filter-caps.
4. Exchanged R3/R13 with 1N4148
5. Added resistor load on output.
Added two 1/2W 680ohm resistors in parallell to each rails output. This will sink approx 44mA at 15V and the resistors will dissipate approx. 0.33W each. Will the resistors do, or do I need power resistors?
Changes to current schematic:
1. Removed the resistors earlier referred to as R13.
2. Changed transistors to the BD139/BD140.
3. Added input filter-caps.
4. Exchanged R3/R13 with 1N4148
5. Added resistor load on output.
Added two 1/2W 680ohm resistors in parallell to each rails output. This will sink approx 44mA at 15V and the resistors will dissipate approx. 0.33W each. Will the resistors do, or do I need power resistors?
Attachments
what I would do
-increase R2/R12 to something like 200-400 Ohms
-place C3/C13 at the base of the transistors
This would make a low pass filter for the noise of the zener
-increase R2/R12 to something like 200-400 Ohms
-place C3/C13 at the base of the transistors
This would make a low pass filter for the noise of the zener
Bricolo said:
Bricolo, I would leave in the base resistors. At switch-off, they prevent the output caps to resverse current through the e-b junction which would ruin the noise performance and Hfe after just one switch-off. Come to think of it, I would put reverse-biased 1N4148's across those e-b junctions to prevent them reverse-biasing at turn off.
Yes, I would follow your advice to put say a few 100 ohms between the zener-resistor junction and the cap to get additional reference filtering. And oh yes Alcaid, turn those 1N4148's around. They should conduct, they are not zeners. Now, if you choose 15v zeners you should get pretty close to 15v, close enough for government work.😉
Jan Didden
till said:
Just a note:
- R5 (10 ohms) increases the output impedance of your supply, meaning ithe output voltage will heavily move all over the place with load changes, lots of impressed ripple. It seems to me that you go to great lengths to get stable stuff at the output emitter and then throw it away with R5.
Jan Didden
One more thing: those 220uF at the output are a short at switch-on. I don't know how those MJE340's react on it, but I would make them smaller, they don't add that much anyway. 50 or 100uF with a good quality foil cap in parallel IMHO would be better. There's no (or not much) danger of oscillations, there's no feedback, it's not a regulator really.
Jan Didden
Jan Didden
Alcaid said:
No one answered my question regarding the resistors on the output. Will two 1/2W metal film in parallell do for each rail? or will they get to hot?
Thanks everyone for all the help on this. Greatly apreciate it. 🙂
Alcaid said:
Come on, give us a break! YOU can calculate the dissipation, can't you? THEN decide if it needs bigger resistors. Take a factor of two for safety if you want.
Jan Didden
How close to the transistor needs the base resistor be to avvoid problems? Need to make some room around the transistor for heatsink.
I know.... lots of stupid questions, but when it comes to electronics, I am kinda stupid
I know.... lots of stupid questions, but when it comes to electronics, I am kinda stupid
As i would understand the solution - let every series resistance/inductance out may be fine for low PS impedance, but gives less ripple or noise filtering.
More RC or LC elemts provide better filtering but higher impedance.
With enough caps behind the filter elemts that add series resitancs, wouldn´t be the AC impedance still very low?
Wouldn´t this mean if there is more filtering, i compensate it with more capacitance behind?
I turn in circles now.
First i was told i need a very clean supply, thus i measured and added components until i had something i could measure very clean. No i´m told i have to make it low z, and so i need to start to remove the elemts i used for filetring...
In post #9 and #10 is to see what kind of noise appears at the IC:
http://www.diyaudio.com/forums/showthread.php?postid=382941#post382941
In 9 we see the spikes coming from spdif interface or so, in #10 we see what happens when the DAC plays 1kHz full scale. Note the 100ns / 1 cm timebase. To me this looks very HFish, shouldn´t this be uninfluenced by series resitance, but should be blocked with caps?
In case there is a cap behing the emitter follower with resistor it should look like a short until it is charged at power on. But what should happen if you take a big enough transistor?
More RC or LC elemts provide better filtering but higher impedance.
With enough caps behind the filter elemts that add series resitancs, wouldn´t be the AC impedance still very low?
Wouldn´t this mean if there is more filtering, i compensate it with more capacitance behind?
I turn in circles now.
First i was told i need a very clean supply, thus i measured and added components until i had something i could measure very clean. No i´m told i have to make it low z, and so i need to start to remove the elemts i used for filetring...
In post #9 and #10 is to see what kind of noise appears at the IC:
http://www.diyaudio.com/forums/showthread.php?postid=382941#post382941
In 9 we see the spikes coming from spdif interface or so, in #10 we see what happens when the DAC plays 1kHz full scale. Note the 100ns / 1 cm timebase. To me this looks very HFish, shouldn´t this be uninfluenced by series resitance, but should be blocked with caps?
In case there is a cap behing the emitter follower with resistor it should look like a short until it is charged at power on. But what should happen if you take a big enough transistor?
mine may not be perfect, but it works. Here is a picture to look at:
http://www.diyaudio.com/forums/showthread.php?postid=382150#post382150
You see left transformers, middle rectifier. Right top zehner regulators with FETs Pass style for +- +0V, Right bottom zehner regulator bipolar +-5V circuit at the moment #27 this thread, playing Saint Saens this moment with the AD1865s.
- Status
- Not open for further replies.