Onvinyl said:
Hi Rüdiger,
The supply i suggested is used for a low gain line amp so a "noisy" zener doesn't matter. Regs for a mc-pre is a different story.
I was also thinking about that low voltage,low noise references but for an open loop regulator it's a bit difficult. Is it posible to put them in series as a normal zener? Say four 5V references for a 20V reg. I never tried this but i suppose the noise is also four times larger but still low. Also the price is relative high.
The local regulator works very fine, try it. John Curl used something like that in his Vendetta amps. His variation is a 220uF bipolar at the gate and a lower resistor(10k or so). I also will try that configuration. But time ...
Somewere in the beginning of the blowtorch thead is a lot of background information about this issue.
Gr.
Johan
Re: serie reg.
1. Swap values on C1 and C2, let the bulk storage do just that and let the regulator remove ripple.
2. Increase LED current, make R1 7.5K instead of 15K, this improves the "knee" and makes the output voltage less susceptible to input voltage variations or ripple.
If the output voltage is 28V, make sure the input voltage is at least 31V (32?) at the bottom of the ripple voltage when under full load. Otherwise the beta of the Darlington will drop too low resulting in poor regulation (noise?). The idea is to have at least 4V across the active regulator/capacitor multiplier Darlington.
Since there is a capacitor at almost every node, the design might oscillate under the right load conditions due to cumulative phase shift.
I haven't built or modeled this design, still I propose some modest schematic changes:joho said:
1. Swap values on C1 and C2, let the bulk storage do just that and let the regulator remove ripple.
2. Increase LED current, make R1 7.5K instead of 15K, this improves the "knee" and makes the output voltage less susceptible to input voltage variations or ripple.
If the output voltage is 28V, make sure the input voltage is at least 31V (32?) at the bottom of the ripple voltage when under full load. Otherwise the beta of the Darlington will drop too low resulting in poor regulation (noise?). The idea is to have at least 4V across the active regulator/capacitor multiplier Darlington.
Since there is a capacitor at almost every node, the design might oscillate under the right load conditions due to cumulative phase shift.
some modest changes
Hi HermanV,
Thank you for the advices.
I use the regulators(pre+local reg) very close to the preamp module. It's like a sandwich. Between the raw dc, which is situated around the transformer, and the regulator is 1m cable. I measued a significent lower noise level at the input of the reg when i use that 100uF (low imp.) caps.
The reg is very stable for voltage variations at the input. 10 volts input "swing" results in a few mv output variation.
The design is more then 10 years old and i don't remember why i used 15k but it's easy to change that value and measure again.
I think the critical input voltage was around 31.5 volt so your're right. i use 35V input.
I have no oscillation problems but the 1N cap between base and collector of Q3/Q4 should reduce such things.
B.t.w. This reg looks very simple and old fashioned but sounds much better than the simple shunt reg in the beginning of this thread.
At this moment i will try some other variations of the last source follower but time...
Johan
Hi HermanV,
Thank you for the advices.
I use the regulators(pre+local reg) very close to the preamp module. It's like a sandwich. Between the raw dc, which is situated around the transformer, and the regulator is 1m cable. I measued a significent lower noise level at the input of the reg when i use that 100uF (low imp.) caps.
The reg is very stable for voltage variations at the input. 10 volts input "swing" results in a few mv output variation.
The design is more then 10 years old and i don't remember why i used 15k but it's easy to change that value and measure again.
I think the critical input voltage was around 31.5 volt so your're right. i use 35V input.
I have no oscillation problems but the 1N cap between base and collector of Q3/Q4 should reduce such things.
B.t.w. This reg looks very simple and old fashioned but sounds much better than the simple shunt reg in the beginning of this thread.
At this moment i will try some other variations of the last source follower but time...
Johan
Hi Maxpou,
Please look at the "UGS adventures" thread first.
http://www.diyaudio.com/forums/showthread.php?postid=1416904#post1416904
The schematic i post in this thread with the BD139/140+source follower and mundorf cap is the one i like best.
It sounds more open than the shunt regulaters i tested. With the shunt regs was more limitation.
http://www.diyaudio.com/forums/showthread.php?postid=1422454#post1422454
Johan
Please look at the "UGS adventures" thread first.
http://www.diyaudio.com/forums/showthread.php?postid=1416904#post1416904
The schematic i post in this thread with the BD139/140+source follower and mundorf cap is the one i like best.
It sounds more open than the shunt regulaters i tested. With the shunt regs was more limitation.
http://www.diyaudio.com/forums/showthread.php?postid=1422454#post1422454
Johan
Super Shunts
A very effective Super Shunt from the late John Linsley Hood, as published originally in E.T.I. This regulator can be configured to work in either polarity, and with suitable component changes has been adapted by "awpagan" from DiyAudio to work with + and - 55V rails in the front end of amplifiers,as wellas currents of >500mA.
A very effective Super Shunt from the late John Linsley Hood, as published originally in E.T.I. This regulator can be configured to work in either polarity, and with suitable component changes has been adapted by "awpagan" from DiyAudio to work with + and - 55V rails in the front end of amplifiers,as wellas currents of >500mA.
An externally hosted image should be here but it was not working when we last tested it.
pietjers said:
Hi Peter,
The answer is no.
I'm working on a second set of supply boards for comparation.
The pcb's are already drilled but i'm very lazy at the moment(voorjaars moeheid).
I also want compare the original UGS serie reg. I ordered the boards and i expect them next month.
Friends of mine use the same regs(without local reg) but it's always a big struggle to find the best match for the last cap(elco).The sound without the local reg can be very good. It's inserted in very expensive systems.
The idee behind the local reg with the suprime cap is to make the
system insensitive for the colouration of that final cap in the pre-reg and in fact you should mount them direct on p.e. preamp boards for the lowest impedance.
Johan
Johan,
The local regulator will add its own color I’m afraid. In audio very often less is more.
Have you ever tried to combine caps (electrolytic + foil caps) or the Black Gate super e-cap as last capacitor?
I like the basic idea of the pre regulator! I took a closer look at the schematic and there are several area’s with possibilities to enhance the level of its performance.
* Replace the led’s by a LM336-2.5 voltage references with let’s say 100uF in parallel. This will increase the performance of the current source with a more stable reference voltage (the zener string) as a result..
* Add a resistor of 22k (to start with) from the emitter of Q1/Q5 to GND to run these transistors a bit more in class A.
* Try a capacitor in parallel with this resistor and listen to the result (let your ears decide).
* Add a third darlington transistor and place both, drivers and pre drivers, in class A with resistors (and caps if that helps). This will give so much current gain that your voltage reference (the zener string) will see nothing from the current demand of the load anymore.
Peter
The local regulator will add its own color I’m afraid. In audio very often less is more.
Have you ever tried to combine caps (electrolytic + foil caps) or the Black Gate super e-cap as last capacitor?
I like the basic idea of the pre regulator! I took a closer look at the schematic and there are several area’s with possibilities to enhance the level of its performance.
* Replace the led’s by a LM336-2.5 voltage references with let’s say 100uF in parallel. This will increase the performance of the current source with a more stable reference voltage (the zener string) as a result..
* Add a resistor of 22k (to start with) from the emitter of Q1/Q5 to GND to run these transistors a bit more in class A.
* Try a capacitor in parallel with this resistor and listen to the result (let your ears decide).
* Add a third darlington transistor and place both, drivers and pre drivers, in class A with resistors (and caps if that helps). This will give so much current gain that your voltage reference (the zener string) will see nothing from the current demand of the load anymore.
Peter
Hi Peter,
Interesting suggestions. The influence of the current(less or more) on the value of the output voltage is a weak point of this regulator(pre+local). It's a open loop reg so you chose for that. Otherwise with a classA preamp it doesn't matter.
I will try some of your suggestions this weekend if i can find some time but my step by step apreach is time absorbing.
The final elcap(englisch?) i use in the prereg is a Panasonic FC.
In the past we try some combinations with foil but it's always a bit dangerous(resonances on high frequencies etc.). The mundorf supreme foil caps are very neutral but don't like elcaps for some reason.
Maybe it's better to start a new thread. It has nothing to do with shuntregs.
B.t.w. what kind of supply do you use?
Johan
Interesting suggestions. The influence of the current(less or more) on the value of the output voltage is a weak point of this regulator(pre+local). It's a open loop reg so you chose for that. Otherwise with a classA preamp it doesn't matter.
I will try some of your suggestions this weekend if i can find some time but my step by step apreach is time absorbing.
The final elcap(englisch?) i use in the prereg is a Panasonic FC.
In the past we try some combinations with foil but it's always a bit dangerous(resonances on high frequencies etc.). The mundorf supreme foil caps are very neutral but don't like elcaps for some reason.
Maybe it's better to start a new thread. It has nothing to do with shuntregs.
B.t.w. what kind of supply do you use?
Johan
Sorry Joho,
i forgot the link in post 28:
http://www.diyaudio.com/forums/showthread.php?postid=945744#post945744
thank you! Maxpou
i forgot the link in post 28:
http://www.diyaudio.com/forums/showthread.php?postid=945744#post945744
thank you! Maxpou
shunt regs
This circuit has no voltage reference. V OUT, is the same as V IN , except for a couple of hundred mV drop through the current limiter transistor. Possibly not the best way to do it, but the normal 1.5A regs seem quite happy when used directly in, bypassing the current regulator.In that case there is no voltage drop. With most loads, the current limiter transistor, only works hard on switch on, and gives a small delay which can often eliminate switch on thumps.
If anybody is interested , send me an email and I will reply with a copy of the original article by JLH. It will be spread over a couple of emails due to size.
SandyK
alexkethel@optusnet.com.au
This circuit has no voltage reference. V OUT, is the same as V IN , except for a couple of hundred mV drop through the current limiter transistor. Possibly not the best way to do it, but the normal 1.5A regs seem quite happy when used directly in, bypassing the current regulator.In that case there is no voltage drop. With most loads, the current limiter transistor, only works hard on switch on, and gives a small delay which can often eliminate switch on thumps.
If anybody is interested , send me an email and I will reply with a copy of the original article by JLH. It will be spread over a couple of emails due to size.
SandyK
alexkethel@optusnet.com.au
Yes, it looks like something went wrong in the schematic.
In a regulating state, Q3's base would sit three Vbe drops above ground, but there is no voltage divider from the output to that node.
If there were 120k from Vout and 10k to GND, it would give some 30 Volts...
EDIT: Saw Sandy's post... now it becomes clear: it's merely a ripple eater circuit...
- Klaus
In a regulating state, Q3's base would sit three Vbe drops above ground, but there is no voltage divider from the output to that node.
If there were 120k from Vout and 10k to GND, it would give some 30 Volts...
EDIT: Saw Sandy's post... now it becomes clear: it's merely a ripple eater circuit...
- Klaus
shunt regs
In the original article, JLH states that his own bench PSU has a noise and ripple , measured over the range 20HZ to 20KHZ of 300uV. Connecting this circuit reduced it to 4uV, with a very low impedance to>100KHZ. C4B is also 2,200uF.
According to JLH , the shunt impedance of the circuit will be <.02 ohms up to some 150KHZ. Replacing C4A and C4B with 4 x 1,000uF 10V low ESR capacitors appears to give even better results, as does 2 x 2,200UF low ESR electros.
SandyK
In the original article, JLH states that his own bench PSU has a noise and ripple , measured over the range 20HZ to 20KHZ of 300uV. Connecting this circuit reduced it to 4uV, with a very low impedance to>100KHZ. C4B is also 2,200uF.
According to JLH , the shunt impedance of the circuit will be <.02 ohms up to some 150KHZ. Replacing C4A and C4B with 4 x 1,000uF 10V low ESR capacitors appears to give even better results, as does 2 x 2,200UF low ESR electros.
SandyK
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.