The best regulator I've built so far
- Thread starter teddy_pardo
- Start date
teddy_pardo
Trade: Teddy Pardo
Almost anything will do. I'm usually using a 50VA transformer and the smoothing capacitors are not critical. Almost anything above 5000uF will do the work, even small snap-in work great.
Agent_Cooper79
pfm Member
I am a bit fresh to the monolithic regs and their implementation so please bear with me.
If one were to use the lm1086 adj (farnell: 9489967), how high rail Voltage can it handle if you are looking for 24 V out (I would like to use Oscons for decoupling). I have measured 35 V rails on my friends Nait3, could I implement this here with that specific reg or do I have to take the less favoured lm317?
Could I use it here if I chose a higher V_out such as 30-ish, what decoupling caps would you then recommend?
Cheers Magnus
If one were to use the lm1086 adj (farnell: 9489967), how high rail Voltage can it handle if you are looking for 24 V out (I would like to use Oscons for decoupling). I have measured 35 V rails on my friends Nait3, could I implement this here with that specific reg or do I have to take the less favoured lm317?
Could I use it here if I chose a higher V_out such as 30-ish, what decoupling caps would you then recommend?
Cheers Magnus
teddy_pardo
Trade: Teddy Pardo
Magnus,
Monolithic regulators such as the 317 or 1086 are only aware of the difference between the output and the input, and this should normally not exceed 35V. As an example you can use 1000V at the input and 965V at the output...
BTW, I'm not sure that the 1086 is better than the 317, it simply has lower dropout. If you look at the datasheet you'll see that on some aspects one is better and on other aspects the other is better. In any case in this regulator it shouldn't make any difference.
Monolithic regulators such as the 317 or 1086 are only aware of the difference between the output and the input, and this should normally not exceed 35V. As an example you can use 1000V at the input and 965V at the output...
BTW, I'm not sure that the 1086 is better than the 317, it simply has lower dropout. If you look at the datasheet you'll see that on some aspects one is better and on other aspects the other is better. In any case in this regulator it shouldn't make any difference.
Agent_Cooper79
pfm Member
Ok, I understand the part about the i/o now. It's just that I got a little bit confused when reading on the farnell site: Voltage, input max:29V.
Thanks for the help, my friend with the Nait3 (my old one) has grown increasingly impatient after hearing my mods
Thanks for the help, my friend with the Nait3 (my old one) has grown increasingly impatient after hearing my mods
BobMaximus
pfm Member
Magnus
You should note that if you short the output of the regulator to ground (0V), then the voltage difference across the regulator is then equal to its *input* voltage. If this voltage is greater than the maximum quoted, then the regulator will be destroyed. I think this is the reason why 29V is quoted (it's the maximum input voltage it can handle when shorted). This is less than the maximum allowable *difference* between the output and the input (ie. 35V). This causes the confusion you have experienced.
Andy
BobMaximus
pfm Member
Mark
The same conclusion would apply to regulating (1) just +ve and (2) both +ve and -ve. The difference is that teddy describes a single-ended supply (+v/0v) and its application to a single-ended pre-amp. I think you are actually asking "would a single-ended pre-amp work better if it were re-designed to be dual railed". This is a different question all together!
Andy
The same conclusion would apply to regulating (1) just +ve and (2) both +ve and -ve. The difference is that teddy describes a single-ended supply (+v/0v) and its application to a single-ended pre-amp. I think you are actually asking "would a single-ended pre-amp work better if it were re-designed to be dual railed". This is a different question all together!
Andy
Mark_III
Active Member
there was a previous thread that had regulation on both legs for a single voltage supply, and I am interested how that compares to this new idea.
e.g.
--------
Vin --------------| 317 |-------- regulated +V
| -------- |
| | R
| --------|
| R
| ----- |
--| 317 |-------------|
----- | |
| R |
| | |
|------- R
| --------|
R | R
| -------- |
0 v --------------| 337 |-------- regulated 0V
--------
or even with a pre-reg on the +ve rail too?
e.g.
--------
Vin --------------| 317 |-------- regulated +V
| -------- |
| | R
| --------|
| R
| ----- |
--| 317 |-------------|
----- | |
| R |
| | |
|------- R
| --------|
R | R
| -------- |
0 v --------------| 337 |-------- regulated 0V
--------
or even with a pre-reg on the +ve rail too?
teddy_pardo
Trade: Teddy Pardo
Mark,
If you refer to the circuit I published about a year or two ago, than this one is much better.
Teddy
If you refer to the circuit I published about a year or two ago, than this one is much better.
Teddy
PigletsDad
My intelligence test came back negative.
teddy,
A wild idea to throw into the mix.
The output impedance of the darlington is controlled by two terms - the drive impedance divided by the current gain + the reciprocal of the transconductance. The second one is basically 26/I ohms for a bipolar, where I is the operating current.
A FET does better on the first term (it has almost infinite current gain), but they are something like 4 to 8 times lower in transconductance at a given current, resulting in higher output impedance.
What about a Fetlington? This is much like a darlington, but uses a fet (a JFET would be good) as the little transistor. The current in the main transistor would be the same, so the transconductance, and hence output impedance would be low, but you get enormous current gain. N channel JFETs have lovely low noise, and are pretty fast.
Just an idea - what do you think?
A wild idea to throw into the mix.
The output impedance of the darlington is controlled by two terms - the drive impedance divided by the current gain + the reciprocal of the transconductance. The second one is basically 26/I ohms for a bipolar, where I is the operating current.
A FET does better on the first term (it has almost infinite current gain), but they are something like 4 to 8 times lower in transconductance at a given current, resulting in higher output impedance.
What about a Fetlington? This is much like a darlington, but uses a fet (a JFET would be good) as the little transistor. The current in the main transistor would be the same, so the transconductance, and hence output impedance would be low, but you get enormous current gain. N channel JFETs have lovely low noise, and are pretty fast.
Just an idea - what do you think?
teddy_pardo
Trade: Teddy Pardo
PD - yes I had this idea in mind too, it sounds like it should be the best, I just don't have a jfet in my spare box... BTW, could you recommend a specific one?
BTW 26/I ohm sounds very high, I measured a total impedance of 3 ohm with a current of 100mA
BTW 26/I ohm sounds very high, I measured a total impedance of 3 ohm with a current of 100mA
PigletsDad
My intelligence test came back negative.
26/I (I in ma), would give 0.26Ohm at 100mA; of course in practice there are other resistances to add, such as source impedance divided by current gain, contact and bond-wire resistances in the transistor and so on.
The 26/I guideline tends to break down at higher currents, but is usually a fair approximation up to a few tens of ma.
The 26/I guideline tends to break down at higher currents, but is usually a fair approximation up to a few tens of ma.
martin clark
pinko bodger
It should be read 26mV/I (mA). 100mA means this component of the output impedance is 0.25ohms.
edit: crosspost with PD!
edit: crosspost with PD!
PigletsDad
My intelligence test came back negative.
There are lots of good JFETS, although not many have a decent voltage rating.
2SK117 would be great if you can get some - I bough a few from Dalbani, but I don't know who else has them. This is a 50V device.
2N3819 is only a 25V part, but readily available.
2SK117 would be great if you can get some - I bough a few from Dalbani, but I don't know who else has them. This is a 50V device.
2N3819 is only a 25V part, but readily available.
I guess some distinguished member of PFM has grabbed the whole lot of tiny boards on Ebay to start production of Teddy's Mega-Regs. Will he ever show up? Is there any other source for this convenient micro Vero board?
By the way, does 2SK117 stand for "Sanken"- the Japanese manufacturer? In that case we´ll have to travel East in quest for them.
Cheers
grabbed the whole lot of tiny boards on Ebay to start production of Teddy's Mega-Regs. Will he ever show up? Is there any other source for this convenient micro Vero board?By the way, does 2SK117 stand for "Sanken"- the Japanese manufacturer? In that case we´ll have to travel East in quest for them.
Cheers
lostintheozone
Cabin Boy
I guess some distinguished member of PFM has
By the way, does 2SK117 stand for "Sanken"- the Japanese manufacturer? In that case we´ll have to travel East in quest for them.
Cheers
BTW if you go to the seller's ebay shop there are more available.
Guy
Seems like 2SK117's are available from here ... http://www.cricklewoodelectronics.com/Cricklewood/product.php?productid=5508&cat=181&page=1
teddy_pardo
Trade: Teddy Pardo
I guess some distinguished member of PFM has
Hey Magnus, you've been spotted...
BTW they have all sorts of PCBs, and they are indeed convenient, once you get use to the concept that instead of cutting strips you have to join them.
