KEEP IN MIND
Hi,
I haven't read the Sprenger article yet but:
Something to keep in mind when picking a CCS is that you need to know exactly how much current you want to regulate and pick your semi-conductor accordingly.
Choosing the wrong device will either starve your tube on current causing extra distortion or the other way around will only function as a current limiter with equally disastrous results.
Often there is no need to go overboard and if you bother to go through specsheets of a number of J-Fets you'll already find quite a number adequate for preamp and small signal service .
Connecting Drain to B+ and Gate and source together to the load will work fine with the devices.
Another aspect to be ware of is that you can't do this with devices working outside class A operation.
Cheers,😉
Hi,
I haven't read the Sprenger article yet but:
Something to keep in mind when picking a CCS is that you need to know exactly how much current you want to regulate and pick your semi-conductor accordingly.
Choosing the wrong device will either starve your tube on current causing extra distortion or the other way around will only function as a current limiter with equally disastrous results.
Often there is no need to go overboard and if you bother to go through specsheets of a number of J-Fets you'll already find quite a number adequate for preamp and small signal service .
Connecting Drain to B+ and Gate and source together to the load will work fine with the devices.
Another aspect to be ware of is that you can't do this with devices working outside class A operation.
Cheers,😉
Be specific, please!
Frank,
So you haven't read it, then you should. I met Emile, he knows exactly and profoundly what he is talking about.
Apart from that, i would be so glad if you spare us this conglomerate of scare talk, generalizing and commonplaces. And be specific instead. Which BJT or FETs worked best for you at which currents?
I find it is obvious, a commonplace, that a BC107 is not the right choice for shunt device of for CCS.
Both CCS transistor and shunt transistor should be capable to handle several times the current the load is spending when idle without getting in trouble with voltage, current, heat dissipation.
A working shuntreg has to waste atleast the same if not twice the current in the shunt device the load is spending. And if the load goes spung!, both CCS and shunt device have to sink that total current without going spung!...
Could you please tell us which J-FETs you used and how your experiences were?
Are you indicating you work without current sensing resistor and rely on the BJT or FET's parameter alone?
All, particular those who use Allen's SuperReg,
i'd like to add some remarks.
1st,
Allen uses a LED for visual control of proper operation: if the reg goes on holiday, the LED is off. Good idea, very good idea. He uses a standard LED capable of 20-30mA. Makes my skin crawl. That this seems to work in most cases is no excuse for my engineering mind, methinks this LED has to survive the same current the shunt device is sinking without a load connected. And this can be way more than 30, even 50 mA. LEDs are not meant to be fuses and while they usually are open circuit when fried, there is no guarantee for it, they can be short-circuit when badly fried. Just take a clear LED and a magnifier and look and the tiny distances inside.
My suggestion: use ceramic-dye SMD LEDs like manufactured by Stettner. www.buerklin.de sells a Stettner CerLED capable of 70mA idle current and >1A of peak current. Use wide routes acting as heasink for the LED might even increase the idle curretn capability and is advisable in any case.
If you run the reg at more than 70mA, please parallel devices and increase the resistor in series with the LED a bit so that the current pretends to spread evenly among the LEDs.
2nd,
Initially the SuperReg came with an LF351 opamp as error amplifier. I had extended discussions about this choice with Allen as in my book the error amplifier is always audible and i had tried fhe LF351 for audio and did not like it.
IN the meantime he uses some fancy Analog Devices opamp, dunno which. And rumours tell that even with this opamp, having a way higher gain-bandwidth product, the circuit does not oscillate. But with the wrong hi-speed opamp it will oscillate. So folks, please get in contact with Allen to learn what the latest state is, what he uses, what he recommends to kill oscillations. I learned from him he is selling those kits in high counts and very few people have problems, most are extremely happy with the SuperReg.
3rd,
the SuperReg is a PD regulator. This means the regulating loop has not only a proportional but also a differentiating component. This differntiating is achieved by C6. C6 shorts R1 for high frequencies.
R1 and R2 (which is a pot) are the sensing voltage devider, R1 is connected to the HV rail. So for slow HV voltage changes the error amplifier sees the fraction of the HV voltage change determined by the sensing voltage devider R1/R2. For fast HV voltage changes the error amplifier sees 100% of the HV voltage change. My recommendation (from theory of course) would be to look very carefully at the size of C6, experiment with it, keep it as small as even possible. Measure performance and listen to it, too. It may be most pleasing under certain circumstances if C6 is missing at all. Oh, and if you use it, it is vital you use caps with outstanding sonics. Not something crappy, no Wima MKP10 or the like.
So,
as you have probably read between the lines, i am not using the SuperReg. At the moment i am listening to a Manfred Huber preamp and Manfred uses his own shunt regulator design; from own exprerience i can tell it sounds just gorgeous. But my decision not to use Allen desing comes from a conceptional decision, not from a sonical one. The preamp design i intend to build has one shunt reg per stage, makes 6 shunt regs for a 3 stage preamp. Allen's SuperReg needs a very clean external power supply for the error opamp. Does it need one supply for 6 regs or does it need 6 supplies?...🙂
My idea is to use 6 regs to eliminate interstage Xtalk and TME, opamps no matter how fancy their CMRR is on the paper, they Xtalk. Yeah, i know, i'm paranoid 🙂, Allen would probably not think 6 supplies are needed.
Manfred's shunt reg design does not need external power for the error amp, it does not even use an opamp, so my insane preamp project becomes a bit less cluttered.
The reg uses a long-tailed pair of BJTs driving a SE BJT driver which drives the shunt device, a FET. This reg also as a PD reg. Equivalent for C6 of Allen's design is C3.
Of course and in alignment what i said above, i am going to get C3 as small as possible. Manfred used 2µ2 and -- as it worked fine on the bench, did not bother further with getting C3 size down.
One thing massively appealing to me is that Manfred's reg has a very small cap (C4) across the output; 2µ2 minimum size allows the use of a coupling-cap-grade foil cap whereas Allens design has an electrolytic of 22µ across the output and he told me that depending on application the needed size can grow up to 100µF and never below 10µF.
And i am a real fan of small caps for power supplies, both AFA physical size and as value is concerned.
So my personal choice is Manfred's design p2p-wired on the terminals of a 5µF or 10µF MKV B25834. Which are sitting on my shelf already.
I want to try out the SuperReg myself one day; i have a manual and description and PCB here.
Frank,
fdegrove said:
So you haven't read it, then you should. I met Emile, he knows exactly and profoundly what he is talking about.
Apart from that, i would be so glad if you spare us this conglomerate of scare talk, generalizing and commonplaces. And be specific instead. Which BJT or FETs worked best for you at which currents?
I find it is obvious, a commonplace, that a BC107 is not the right choice for shunt device of for CCS.
Both CCS transistor and shunt transistor should be capable to handle several times the current the load is spending when idle without getting in trouble with voltage, current, heat dissipation.
A working shuntreg has to waste atleast the same if not twice the current in the shunt device the load is spending. And if the load goes spung!, both CCS and shunt device have to sink that total current without going spung!...
Could you please tell us which J-FETs you used and how your experiences were?
Are you indicating you work without current sensing resistor and rely on the BJT or FET's parameter alone?
This makes sense to me as both CCS transistor and shunt device are single ended amplifier topologies. But then, without the semiconductors being at a reasonable operating point neither CCS nor shunt device would cooperate as a working regulator.
All, particular those who use Allen's SuperReg,
i'd like to add some remarks.
1st,
Allen uses a LED for visual control of proper operation: if the reg goes on holiday, the LED is off. Good idea, very good idea. He uses a standard LED capable of 20-30mA. Makes my skin crawl. That this seems to work in most cases is no excuse for my engineering mind, methinks this LED has to survive the same current the shunt device is sinking without a load connected. And this can be way more than 30, even 50 mA.
LEDs are not meant to be fuses and while they usually are open circuit when fried, there is no guarantee for it, they can be short-circuit when badly fried. Just take a clear LED and a magnifier and look and the tiny distances inside.My suggestion: use ceramic-dye SMD LEDs like manufactured by Stettner. www.buerklin.de sells a Stettner CerLED capable of 70mA idle current and >1A of peak current. Use wide routes acting as heasink for the LED might even increase the idle curretn capability and is advisable in any case.
If you run the reg at more than 70mA, please parallel devices and increase the resistor in series with the LED a bit so that the current pretends to spread evenly among the LEDs.
2nd,
Initially the SuperReg came with an LF351 opamp as error amplifier. I had extended discussions about this choice with Allen as in my book the error amplifier is always audible and i had tried fhe LF351 for audio and did not like it.
IN the meantime he uses some fancy Analog Devices opamp, dunno which. And rumours tell that even with this opamp, having a way higher gain-bandwidth product, the circuit does not oscillate. But with the wrong hi-speed opamp it will oscillate. So folks, please get in contact with Allen to learn what the latest state is, what he uses, what he recommends to kill oscillations. I learned from him he is selling those kits in high counts and very few people have problems, most are extremely happy with the SuperReg.
3rd,
the SuperReg is a PD regulator. This means the regulating loop has not only a proportional but also a differentiating component. This differntiating is achieved by C6. C6 shorts R1 for high frequencies.
R1 and R2 (which is a pot) are the sensing voltage devider, R1 is connected to the HV rail. So for slow HV voltage changes the error amplifier sees the fraction of the HV voltage change determined by the sensing voltage devider R1/R2. For fast HV voltage changes the error amplifier sees 100% of the HV voltage change. My recommendation (from theory of course) would be to look very carefully at the size of C6, experiment with it, keep it as small as even possible. Measure performance and listen to it, too. It may be most pleasing under certain circumstances if C6 is missing at all. Oh, and if you use it, it is vital you use caps with outstanding sonics. Not something crappy, no Wima MKP10 or the like.
So,
as you have probably read between the lines, i am not using the SuperReg. At the moment i am listening to a Manfred Huber preamp and Manfred uses his own shunt regulator design; from own exprerience i can tell it sounds just gorgeous. But my decision not to use Allen desing comes from a conceptional decision, not from a sonical one. The preamp design i intend to build has one shunt reg per stage, makes 6 shunt regs for a 3 stage preamp. Allen's SuperReg needs a very clean external power supply for the error opamp. Does it need one supply for 6 regs or does it need 6 supplies?
...🙂My idea is to use 6 regs to eliminate interstage Xtalk and TME, opamps no matter how fancy their CMRR is on the paper, they Xtalk. Yeah, i know, i'm paranoid 🙂, Allen would probably not think 6 supplies are needed.
Manfred's shunt reg design does not need external power for the error amp, it does not even use an opamp, so my insane preamp project becomes a bit less cluttered.
The reg uses a long-tailed pair of BJTs driving a SE BJT driver which drives the shunt device, a FET. This reg also as a PD reg. Equivalent for C6 of Allen's design is C3.
Of course and in alignment what i said above, i am going to get C3 as small as possible. Manfred used 2µ2 and -- as it worked fine on the bench, did not bother further with getting C3 size down.
One thing massively appealing to me is that Manfred's reg has a very small cap (C4) across the output; 2µ2 minimum size allows the use of a coupling-cap-grade foil cap whereas Allens design has an electrolytic of 22µ across the output and he told me that depending on application the needed size can grow up to 100µF and never below 10µF.
And i am a real fan of small caps for power supplies, both AFA physical size and as value is concerned.
So my personal choice is Manfred's design p2p-wired on the terminals of a 5µF or 10µF MKV B25834. Which are sitting on my shelf already.
I want to try out the SuperReg myself one day; i have a manual and description and PCB here.
CCS.
Hi,
To reply to part of your Q:
Most of these FETs I have experience with are low current devices and are always preceded by another semi-conductor current sensing device arrangement.
I have experienced the BF range and MPSA combos.
It is painting with a wide brush if you're not being specific about the task to perform though.
Yes,it is a waste of energy but not as much as you point out.
As I said there is no point in applying a CCS for a non defined load and as such I can only repeat: CCSs for Class A operation only please.
Don't even try to current or voltage regulate anything that goes deeply into Class B ,only ample current reserves will help here.
A CCS would only be a rope around the neck of such a stage.
All in all a current reg works along the same line as a voltage reg.,only the comparator/error amp. has to react to current changes not to voltage changes and this may be new to most:
FROM THE PSU,NOT FROM THE CIRCUIT IT SELF.
Ideally the voltage and current consumption of a Class A design should be held constant and the regulation should work as an unobtrusive trickle charger.
The operative circuit should never see the charger work:infinite isolation.
I fact all the problems one may encounter you seem to understand so there is no need for me to shoot down someones' business here.
It has been pointed out before by me and I'm prety much convinced you see my point.
Hi,
To reply to part of your Q:
Most of these FETs I have experience with are low current devices and are always preceded by another semi-conductor current sensing device arrangement.
I have experienced the BF range and MPSA combos.
It is painting with a wide brush if you're not being specific about the task to perform though.
Yes,it is a waste of energy but not as much as you point out.
As I said there is no point in applying a CCS for a non defined load and as such I can only repeat: CCSs for Class A operation only please.
Don't even try to current or voltage regulate anything that goes deeply into Class B ,only ample current reserves will help here.
A CCS would only be a rope around the neck of such a stage.
All in all a current reg works along the same line as a voltage reg.,only the comparator/error amp. has to react to current changes not to voltage changes and this may be new to most:
FROM THE PSU,NOT FROM THE CIRCUIT IT SELF.
Ideally the voltage and current consumption of a Class A design should be held constant and the regulation should work as an unobtrusive trickle charger.
The operative circuit should never see the charger work:infinite isolation.
I fact all the problems one may encounter you seem to understand so there is no need for me to shoot down someones' business here.
It has been pointed out before by me and I'm prety much convinced you see my point.
Parts source for M.Hubers P.S. regulator?
Does anyone know where I can get the parts for Manfred Hubers regulator? I can't seem to get them anywhere i usually get stuff (mouser, digikey, mcm).
These are some of the parts i can't find:
ZPD10 (zener)
BSS135 (mosfet)?
MTP2P5O
ZTX458
ZTX558
He mentioned farnell.com, but that doesn't seem to work.
Thanks for any info!
Matt Mitchell
Does anyone know where I can get the parts for Manfred Hubers regulator? I can't seem to get them anywhere i usually get stuff (mouser, digikey, mcm).
These are some of the parts i can't find:
ZPD10 (zener)
BSS135 (mosfet)?
MTP2P5O
ZTX458
ZTX558
He mentioned farnell.com, but that doesn't seem to work.
Thanks for any info!
Matt Mitchell
-------------------------------NickC said:
Simplest one is in the TL431 data sheet. Shunt this with a low esr largish cap, and you can get a 10 uV noise supply with 50 mA and more depending on the p transistor you choose. Pay attention to using a lowish current thru'the TL431.
Emile Sprenger's Shunt reg is there for all to use apparently.
He has a manual, parts list schematic...the works on a website.
http://www.euronet.nl/~aespreng/Shuntregv11d/
I believe it is state of the art as far as shuntregs go...
Cheers,
Bas
He has a manual, parts list schematic...the works on a website.
http://www.euronet.nl/~aespreng/Shuntregv11d/
I believe it is state of the art as far as shuntregs go...
Cheers,
Bas
.About Wright
I am not surprised that Allen's operation is a single man one!
From my 5 min experience with the guy the association of Mr. Wright with a first class, genuine A-hole is not a very hard one to make.
Imbued with the hype from opinions of the large number of people that have listened to the products form this sound 'guru', I was ready to purchase his pp tube amp.
I asked him some price info about his pp tube amp and what I got in return was a rude reply as to the extent that I shouldn't bug him, that the tube amp wasn't ready and even if it was it wouldn't have been for sale in the US or whatever.
Although I don't have my original email, I can say that my inquiry was a serious one, since I had pretty much made up my mind.
According to the website the damn thing was ready for shipping.
I wasn't too kind in my reply to him.
For someone claiming to make a living out of the hi end audio business that's a pretty spunky attitude considering the number of client and the fierce competition.
My guess some people are mesmerized by bad attitude, they consider it a sign of knowledge and confer the person some type of status.
I am not surprised that Allen's operation is a single man one!
From my 5 min experience with the guy the association of Mr. Wright with a first class, genuine A-hole is not a very hard one to make.
Imbued with the hype from opinions of the large number of people that have listened to the products form this sound 'guru', I was ready to purchase his pp tube amp.
I asked him some price info about his pp tube amp and what I got in return was a rude reply as to the extent that I shouldn't bug him, that the tube amp wasn't ready and even if it was it wouldn't have been for sale in the US or whatever.
Although I don't have my original email, I can say that my inquiry was a serious one, since I had pretty much made up my mind.
According to the website the damn thing was ready for shipping.
I wasn't too kind in my reply to him.
For someone claiming to make a living out of the hi end audio business that's a pretty spunky attitude considering the number of client and the fierce competition.
My guess some people are mesmerized by bad attitude, they consider it a sign of knowledge and confer the person some type of status.
I guess that would make them a GuRude
"My guess some people are mesmerized by bad attitude, they consider it a sign of knowledge and confer the person some type of status."
I wouldn't put money on that bet if I were you ...........
Besides I am nice to people who are going to give me money. Why you would not believe how nice I can be and for how little money........ Why, it's one of the best shopping bargins out there!
"That this seems to work in most cases is no excuse for my engineering mind, methinks this LED has to survive the same current the shunt device is sinking without a load connected"
Put a resistor in parallel with the LED. Pick the value that gives the voltage that you would measure with 5mA across the LED. You can test the LED first at this current to find this voltage. That was so easy that I won't even ask for money.
"My guess some people are mesmerized by bad attitude, they consider it a sign of knowledge and confer the person some type of status."
I wouldn't put money on that bet if I were you ...........
Besides I am nice to people who are going to give me money. Why you would not believe how nice I can be and for how little money........ Why, it's one of the best shopping bargins out there!
"That this seems to work in most cases is no excuse for my engineering mind, methinks this LED has to survive the same current the shunt device is sinking without a load connected"
Put a resistor in parallel with the LED. Pick the value that gives the voltage that you would measure with 5mA across the LED. You can test the LED first at this current to find this voltage. That was so easy that I won't even ask for money.
State of the art
Maybe not........ I think a better design could be done with 70 to 80% the parts count. there are some things in the design that are pretty sloppy as well. I will wait for the stones, bricks, bottles, and curses to be thrown at me and then will elaborate......... Let he who is without Spice throw the first stone.
Maybe not........ I think a better design could be done with 70 to 80% the parts count. there are some things in the design that are pretty sloppy as well. I will wait for the stones, bricks, bottles, and curses to be thrown at me and then will elaborate......... Let he who is without Spice throw the first stone.
Hi,
Since I am literally without Spice I'll throw the stone...make that the gauntlet, your way and cordially invite you to give us whatever you would like to get of your chest.
After all it won't hurt anyone if we could somehow devise a better CCS, would it?
Cheers,😉
Since I am literally without Spice I'll throw the stone...make that the gauntlet, your way and cordially invite you to give us whatever you would like to get of your chest.
After all it won't hurt anyone if we could somehow devise a better CCS, would it?
Cheers,😉
We're both regulator guys having a nice regulator discussion
x
I wonder at the use of LT1085 regulator for the supply for error amp. Why not a shunt regulator it as well with 400 or 500 volts, even a resistor bias shunt element would give very good rejection for the error amp. The current preloading is on the low side and the use of the voltage setting resistors as the current load instead of a separate load resistor causes a real compromise in the PSRR that larger resistors with the same 15 uF cap would give.
Biasing the U3 reference with the adjustment current from LT1085 would be even better and would improve the PSRR and lower the output noise from the LT1085 and lessen parts count.
The current source T2 is regulated and with the TL491 as the error amp. I don't know if this is as good sounding as the IRF810 would be with a more simple source resistor only feedback scheme.
The mosfet T1 current source for the reference error amp U2 needs a gate resistor. A series resistor at the drain of a 1K to 5K resistor would filter RF out from the error amp U2 better.
There is needs to be a RC filter between the voltage reference U3 and the 200 ohm error amp feedback resistor. There is too much noise as the design sits now.
The TL491 is a noisy reference and the noise is not even specified by several manufacturers! (not good, very discouraging) When set for 6.9 volts it is not even as good as the LM329 from one the one vendor I saw who would characterize the noise. The are much better references than either of these now available.
Most of the suggestions on the LT1085 were standard tweaks from the Audio Amateur at least a decade ago and the rational and performance improvements can be verified by reading the applications part of the data sheet. It is a very interesting design with some very clever circuits and uses first rate parts for most of the design. Having a great interest in power supplies for audio I can honestly say that the LT1085 part of the circuit was not state of the art 10 years ago much less now.......
I will save the main error amp stuff for later and there are several suggestions for that as well. It would be interesting to see Andy's (AWL) comments as well.
x
I wonder at the use of LT1085 regulator for the supply for error amp. Why not a shunt regulator it as well with 400 or 500 volts, even a resistor bias shunt element would give very good rejection for the error amp. The current preloading is on the low side and the use of the voltage setting resistors as the current load instead of a separate load resistor causes a real compromise in the PSRR that larger resistors with the same 15 uF cap would give.
Biasing the U3 reference with the adjustment current from LT1085 would be even better and would improve the PSRR and lower the output noise from the LT1085 and lessen parts count.
The current source T2 is regulated and with the TL491 as the error amp. I don't know if this is as good sounding as the IRF810 would be with a more simple source resistor only feedback scheme.
The mosfet T1 current source for the reference error amp U2 needs a gate resistor. A series resistor at the drain of a 1K to 5K resistor would filter RF out from the error amp U2 better.
There is needs to be a RC filter between the voltage reference U3 and the 200 ohm error amp feedback resistor. There is too much noise as the design sits now.
The TL491 is a noisy reference and the noise is not even specified by several manufacturers! (not good, very discouraging) When set for 6.9 volts it is not even as good as the LM329 from one the one vendor I saw who would characterize the noise. The are much better references than either of these now available.
Most of the suggestions on the LT1085 were standard tweaks from the Audio Amateur at least a decade ago and the rational and performance improvements can be verified by reading the applications part of the data sheet. It is a very interesting design with some very clever circuits and uses first rate parts for most of the design. Having a great interest in power supplies for audio I can honestly say that the LT1085 part of the circuit was not state of the art 10 years ago much less now.......
I will save the main error amp stuff for later and there are several suggestions for that as well. It would be interesting to see Andy's (AWL) comments as well.
But I still haven't found what I'm looking for
Not the U2, I was refering to but a great album and the first CD I bought.
Sorry about the lost links
http://www.euronet.nl/~aespreng/Shuntregv11d/Shuntreg11d-sch.pdf
http://www.euronet.nl/~aespreng/Shuntregv11d/
Not the U2, I was refering to but a great album and the first CD I bought.
Sorry about the lost links
http://www.euronet.nl/~aespreng/Shuntregv11d/Shuntreg11d-sch.pdf
http://www.euronet.nl/~aespreng/Shuntregv11d/
Attachments
Re: We're both regulator guys having a nice regulator discussion
Ok, I guess I have to answer this one.
Since yesterday I'm a member of this forum (I had to join in order to reply).
Normally I'm over at the Joelist .. not enough time for 2 lists ..
As for the shuntreg .. it's been a while since I worked on the design .. funny that there;s a discussion on now .. the reg's have been running for over a year now in my preamp .. very, very nice ..
I'm really pleased with the comments made. Some are new, others I realised after the design (maybe something for the next version?)
Let me start by saying that I'm certain improvements can be made .. but .. as the reg is now, it beats anything I've heard so far.
>> I wonder at the use of LT1085 regulator for the supply for error amp.
Simple, not so bad regulator .. remember we're not talking about regulating a phono amp (direct signal path), but the error amp of the regulator of a tube amp .. so I chose a simple solution.
No, the 1085 is not a state of the art regulator, but it's so far down from the signal path, I don't think its much of a concern.
The PSRR of the 655 is pretty good as well (65/70dB) .. some room for improvement, nevertheless.
>> Why not a shunt regulator it as well with 400 or 500 volts, even a resistor bias shunt element would give very good rejection for the error amp.
I take it you want to power the opamp from the B+ .. not so good idea due to the large waste (on the regulated side it would never work anyway) IIRC, the error amp draws around 20mA. @ 400-500V that's a lot of power to dissipate. Not the route I would choose. No if you want to improve things, go for a better low V regulator.
>> The current preloading is on the low side
I tried higher currents, but could not measure/hear any difference (with a HP spectrum analyzer and >100dB horns), so why raise the current? Even with the substantial Cin of the IRF, the output swing of the opamp is very small, so there's no chance of slewing the output.
>> the use of the voltage setting resistors as the current load instead of a separate load resistor causes a real compromise in the PSRR that larger resistors with the same 15 uF cap would give
Agreed. The voltage ref can certainly be improved (noise). The PSRR is not that bad due to the current source feeding the 431.
>> Biasing the U3 reference with the adjustment current from LT1085 would be even better and would improve the PSRR and lower the output noise from the LT1085 and lessen parts count.
Never thougth of that .. smart thinking.
>> The current source T2 is regulated and with the TL491 as the error amp. I don't know if this is as good sounding as the IRF810 would be with a more simple source resistor only feedback scheme.
I tried both .. the 431 error amp gave a subjective blacker background (compared to a simple MOSFET/source resistor CCS)
BTW its a TL431 not a TL491
>> The mosfet T1 current source for the reference error amp U2 needs a gate resistor. A series resistor at the drain of a 1K to 5K resistor would filter RF out from the error amp U2 better.
Agreed .. although I haven't had any RF problems with the reg (the separate groundplane for the CCS probably helps).
>>There is needs to be a RC filter between the voltage reference U3 and the 200 ohm error amp feedback resistor. There is too much noise as the design sits now.
I thought aboput this one during the design. I chose not to do this due to the introduction of another pole. As the circuit is now, noise is very, very low. With a regulated B+ of 400V there's about 200-300uV of noise ..
>> TL491 is a noisy reference and the noise is not even specified by several manufacturers! (not good, very discouraging)
All the datasheets I have show 40nV/HZ^-2 .. I've seen better, but it's not as bad as you suggest.
>> When set for 6.9 volts it is not even as good as the LM329 from one the one vendor I saw who would characterize the noise. The are much better references than either of these now available.
The datasheets list the 329 at 75nV/Hz^-2 .. that's amost 2* the noise of the 431 ..
>> Most of the suggestions on the LT1085 were standard tweaks from the Audio Amateur at least a decade ago and the rational and performance improvements can be verified by reading the applications part of the data sheet.
As I said before, due to the PSRR of the 655, I didn't put too much effort in the lt1085 reg .. I realise that this aspect could be improved ..
However .. will it have any sonic impact?
>> It is a very interesting design with some very clever circuits and uses first rate parts for most of the design.
Yup
>> Having a great interest in power supplies for audio I can honestly say that the LT1085 part of the circuit was not state of the art 10 years ago much less now.......
As I said before, the LT1085 only being used to power the error amp of the regulator of a tube circuit. Very far removed from the signal ..
>> I will save the main error amp stuff for later and there are several suggestions for that as well.
Looking forward to hearing more.
Emile
Ok, I guess I have to answer this one.
Since yesterday I'm a member of this forum (I had to join in order to reply).
Normally I'm over at the Joelist .. not enough time for 2 lists ..
As for the shuntreg .. it's been a while since I worked on the design .. funny that there;s a discussion on now .. the reg's have been running for over a year now in my preamp .. very, very nice ..
I'm really pleased with the comments made. Some are new, others I realised after the design (maybe something for the next version?)
Let me start by saying that I'm certain improvements can be made .. but .. as the reg is now, it beats anything I've heard so far.
>> I wonder at the use of LT1085 regulator for the supply for error amp.
Simple, not so bad regulator .. remember we're not talking about regulating a phono amp (direct signal path), but the error amp of the regulator of a tube amp .. so I chose a simple solution.
No, the 1085 is not a state of the art regulator, but it's so far down from the signal path, I don't think its much of a concern.
The PSRR of the 655 is pretty good as well (65/70dB) .. some room for improvement, nevertheless.
>> Why not a shunt regulator it as well with 400 or 500 volts, even a resistor bias shunt element would give very good rejection for the error amp.
I take it you want to power the opamp from the B+ .. not so good idea due to the large waste (on the regulated side it would never work anyway) IIRC, the error amp draws around 20mA. @ 400-500V that's a lot of power to dissipate. Not the route I would choose. No if you want to improve things, go for a better low V regulator.
>> The current preloading is on the low side
I tried higher currents, but could not measure/hear any difference (with a HP spectrum analyzer and >100dB horns), so why raise the current? Even with the substantial Cin of the IRF, the output swing of the opamp is very small, so there's no chance of slewing the output.
>> the use of the voltage setting resistors as the current load instead of a separate load resistor causes a real compromise in the PSRR that larger resistors with the same 15 uF cap would give
Agreed. The voltage ref can certainly be improved (noise). The PSRR is not that bad due to the current source feeding the 431.
>> Biasing the U3 reference with the adjustment current from LT1085 would be even better and would improve the PSRR and lower the output noise from the LT1085 and lessen parts count.
Never thougth of that .. smart thinking.
>> The current source T2 is regulated and with the TL491 as the error amp. I don't know if this is as good sounding as the IRF810 would be with a more simple source resistor only feedback scheme.
I tried both .. the 431 error amp gave a subjective blacker background (compared to a simple MOSFET/source resistor CCS)
BTW its a TL431 not a TL491
>> The mosfet T1 current source for the reference error amp U2 needs a gate resistor. A series resistor at the drain of a 1K to 5K resistor would filter RF out from the error amp U2 better.
Agreed .. although I haven't had any RF problems with the reg (the separate groundplane for the CCS probably helps).
>>There is needs to be a RC filter between the voltage reference U3 and the 200 ohm error amp feedback resistor. There is too much noise as the design sits now.
I thought aboput this one during the design. I chose not to do this due to the introduction of another pole. As the circuit is now, noise is very, very low. With a regulated B+ of 400V there's about 200-300uV of noise ..
>> TL491 is a noisy reference and the noise is not even specified by several manufacturers! (not good, very discouraging)
All the datasheets I have show 40nV/HZ^-2 .. I've seen better, but it's not as bad as you suggest.
>> When set for 6.9 volts it is not even as good as the LM329 from one the one vendor I saw who would characterize the noise. The are much better references than either of these now available.
The datasheets list the 329 at 75nV/Hz^-2 .. that's amost 2* the noise of the 431 ..
>> Most of the suggestions on the LT1085 were standard tweaks from the Audio Amateur at least a decade ago and the rational and performance improvements can be verified by reading the applications part of the data sheet.
As I said before, due to the PSRR of the 655, I didn't put too much effort in the lt1085 reg .. I realise that this aspect could be improved ..
However .. will it have any sonic impact?
>> It is a very interesting design with some very clever circuits and uses first rate parts for most of the design.
Yup
>> Having a great interest in power supplies for audio I can honestly say that the LT1085 part of the circuit was not state of the art 10 years ago much less now.......
As I said before, the LT1085 only being used to power the error amp of the regulator of a tube circuit. Very far removed from the signal ..
>> I will save the main error amp stuff for later and there are several suggestions for that as well.
Looking forward to hearing more.
Emile
Count me among the people who are subsidizing AW's lifestyle and receiving no product (and no answer to e-mails) after more than four months. No "SuperCables" book can be worth this type of treatment.
I have requested a refund and a response. If He can't bother to do that (he certainly knows how to take people's money efficiently, I'll give him that) I am going to file fraud papers with Visa.
This is a crock.
IMO, a person would have to me a massochist to order anything from him. He is ****ing all over the DIY community. Shame on him.
GnD
I have requested a refund and a response. If He can't bother to do that (he certainly knows how to take people's money efficiently, I'll give him that) I am going to file fraud papers with Visa.
This is a crock.
IMO, a person would have to me a massochist to order anything from him. He is ****ing all over the DIY community. Shame on him.
GnD
Look, Graham
some putz audiophile from some magazine or website (can't recall which) said to have paid him 5 grands ($5,000) as a down payment for some tube amps and he had to wait a couple of years to get a prototype for the VSAC2003.
How much did you pay him for that book? 50 bucks?
You do the math how long is going to take you to get the book 😉
I guess AW has trascended this universe and he lives in some Star Trek parallel reality were money (and ethics) are of no importance. I borrowed the preamp cookbook from a friend, it's the first edition. It's about 2 pages worth of schematics and 40 pages of BS about when he was jerking this guy off that he referred to as his 'guru'.
Bottom line: get your money back.
some putz audiophile from some magazine or website (can't recall which) said to have paid him 5 grands ($5,000) as a down payment for some tube amps and he had to wait a couple of years to get a prototype for the VSAC2003.
How much did you pay him for that book? 50 bucks?
You do the math how long is going to take you to get the book 😉
I guess AW has trascended this universe and he lives in some Star Trek parallel reality were money (and ethics) are of no importance. I borrowed the preamp cookbook from a friend, it's the first edition. It's about 2 pages worth of schematics and 40 pages of BS about when he was jerking this guy off that he referred to as his 'guru'.
Bottom line: get your money back.
I've never had any dealings with Allan so I can't comment on any of that. But it does sound a shame that a few people have had some trouble recieving what they've paid for, especially given that some of his work (eg this thread, SLCF, etc.) sets an insightful standard for the DIY community. (It was also good of him to seemingly post a thoughful reply here.)
Given that many of us here at the DIYboard will require the use of a decent a current source of some kind for our own circuit designs, and given that one of the argueably sonically good designs (Allan's) could be improved still further; then how about we try and evolve this thread into some kid of a "foundation board prototype" of one of our own?
Maestro Dieckmann has offered some good suggestions, let's keep the ball rolling...?
Given that many of us here at the DIYboard will require the use of a decent a current source of some kind for our own circuit designs, and given that one of the argueably sonically good designs (Allan's) could be improved still further; then how about we try and evolve this thread into some kid of a "foundation board prototype" of one of our own?
Maestro Dieckmann has offered some good suggestions, let's keep the ball rolling...?
- Status
- Not open for further replies.