Bill, what I see looks great. As for using polystyrenes on the diodes, they should be fine if the ratings are sufficient. I've had small polystyrenes on the vector impedance meter, and they're good to way high MHz, far better than I thought they'd be. Q is extremely high, and I don't know if damping is needed for diode snubber use- I never do and it seems fine.
IMO, the three locations have a far better chance of affecting something than before the reg. Being an impossible skeptic, I still have to ask what's downstream of the caps. Until I got my preamp circuitry truly properly bypassed, I could affect it with the supply. Once I had local film bypasses on the op-amps, nothing I did with the supply made any significant difference. The only other thing I'd control, and it's probably being paranoid, is the value of the caps being tested. Get 'em on a bridge and record the true values for posterity. I don't like to reach conclusions about caps that aren't really the same. OTOH, I find it hard to believe a circuit would care about the difference between 3000 uF and 2400 uF, a 20% tolerance difference.
IMO, the three locations have a far better chance of affecting something than before the reg. Being an impossible skeptic, I still have to ask what's downstream of the caps. Until I got my preamp circuitry truly properly bypassed, I could affect it with the supply. Once I had local film bypasses on the op-amps, nothing I did with the supply made any significant difference. The only other thing I'd control, and it's probably being paranoid, is the value of the caps being tested. Get 'em on a bridge and record the true values for posterity. I don't like to reach conclusions about caps that aren't really the same. OTOH, I find it hard to believe a circuit would care about the difference between 3000 uF and 2400 uF, a 20% tolerance difference.
I think ceramic is excellent, but only before the regulator.
It seems the new Panasonic FM is the same as Panasonic FC but with lower impedance at high frequencies. I am not sure if that is what I want.
What I really want is a flat curve in ESR/ESL without major nasty internal resonances.
The cap at the adjustment pin is the most sensitive.
I think I will buy a pair of Panasonic 470uF/25v in the coming weekend to test them, in replacement of the Xicon.
I have a plenty of Rubyzon ZL but I don't like them anymore.
It seems the new Panasonic FM is the same as Panasonic FC but with lower impedance at high frequencies. I am not sure if that is what I want.
What I really want is a flat curve in ESR/ESL without major nasty internal resonances.
The cap at the adjustment pin is the most sensitive.
I think I will buy a pair of Panasonic 470uF/25v in the coming weekend to test them, in replacement of the Xicon.
I have a plenty of Rubyzon ZL but I don't like them anymore.
The stage that is powered has 6 dual opamps. Each opamp has its direct, separate rail and ground paths to the 4,700uF capacitor. A Rubycon ZL 120uF/25v is placed within 1-2cm from each opamp on each rail, and in addition, a 0.01uF polystyrene is soldered at the back of the PCB directly on the opamp power supply pins. This is the bypassing I use. So there are 12 Rubycon ZL 100uF and 12 polystyrene 0.01uF used for the local bypass.
I initially used 0.1uF polypropylene at the supply pins of the opamps. This is according to the datasheets of the opa627 opamps. However, because the Rubycon ZL has a fairly low impedance already, paralleling the bypass caps created a dip in the impedance somewhere at a high frequency possibly within the audioband. The treble was grossly exaggerated. So I replaced the 0.1uF MKP with the 0.01uF polystyrene. The exaggeration at high frequency was reduced to an acceptable level.
I found if the local bypass already uses the Rubycon ZL, and if the power supply again uses the Rubycon ZL, the treble becomes too exaggerated and quite messy.
I initially used 0.1uF polypropylene at the supply pins of the opamps. This is according to the datasheets of the opa627 opamps. However, because the Rubycon ZL has a fairly low impedance already, paralleling the bypass caps created a dip in the impedance somewhere at a high frequency possibly within the audioband. The treble was grossly exaggerated. So I replaced the 0.1uF MKP with the 0.01uF polystyrene. The exaggeration at high frequency was reduced to an acceptable level.
I found if the local bypass already uses the Rubycon ZL, and if the power supply again uses the Rubycon ZL, the treble becomes too exaggerated and quite messy.
Bill, some time ago in a long forgotten thread, I pondered this esr business- what would it take to maintain some arbitrary low impedance over the full audio band? It turns out that if you pick some value, say Z=1 ohm, it's pretty easy at the low end with big electrolytics, and pretty easy at the high end with just about any film or ceramic, but very difficult in the middle. You need a combination of fairly high value combined with a lower esr/esl than electrolytics have. The solution (that I didn't try) seemed to be large poly motor run capacitors. Try it. Choose some value for Z that makes you happy and see what combination of caps will insure it over the full audio band 
Conrad,
I agree with many things you said. But did you mean that the impedance / esr at mid frequencies can be higher than at low frequencies? I thought that above Fc, impedance due to inductance will go higher and higher while esr remains constant. I guess that when manufacturer refers to impedance, they include both the inductance and esr. Correct me if I am wrong. I am not sure if I have the space in my line level stuff to house any motor run caps. I will live with electrolytic and film caps.
Funny I searched the Farnell and RS Components catalogues and found only low esl capacitors and not low esr capacitors!
Manufacturers usually only have the esl at a single frequency, such as 0.015R at 100kHz, but never shows the ESR.
I wonder if we can work it out from the maximum ripple current factors that manufacturers usually give from 100Hz to 100kHz. Can we assume that at 0.5 the current the impedance is twice?
By the way, what is the typical ESR for a large capacitor, say, 4,700uF / 25v?
Regards,
Bill
I agree with many things you said. But did you mean that the impedance / esr at mid frequencies can be higher than at low frequencies? I thought that above Fc, impedance due to inductance will go higher and higher while esr remains constant. I guess that when manufacturer refers to impedance, they include both the inductance and esr. Correct me if I am wrong. I am not sure if I have the space in my line level stuff to house any motor run caps. I will live with electrolytic and film caps.
Funny I searched the Farnell and RS Components catalogues and found only low esl capacitors and not low esr capacitors!
Manufacturers usually only have the esl at a single frequency, such as 0.015R at 100kHz, but never shows the ESR.
I wonder if we can work it out from the maximum ripple current factors that manufacturers usually give from 100Hz to 100kHz. Can we assume that at 0.5 the current the impedance is twice?
By the way, what is the typical ESR for a large capacitor, say, 4,700uF / 25v?
Regards,
Bill
comparing modern electrolytic caps
Conrad
Did you get a chance to try that JLH "ripple eater" ? It overcomes these capacitor problems, AND seems to improve dynamics, as well as a very impressive bottom end. It is being used by many members with a modified Silicon Chip Headphone Amplifier design, using kits from Jaycar and Altronics. (Australia)
Many members in Rock Grotto are now reporting fabulous results when it uses 2x 2,200uF low ESR in parallel , where JLH used 2 x ordinary electros.Even better, is the JLH when the current limiter section is bypassed, and the unit connected directly to 78xx/79xx or LM317/LM337 . The initial high switch on surge is handled by the current limiting in the VREGs.We are now using a smaller dual polarity PCB, which due to member interest after impressive posted results, is being supplied to members for little more than cost, by an ex Broadcast engineer, now lecturer.
SandyK
Conrad
Did you get a chance to try that JLH "ripple eater" ? It overcomes these capacitor problems, AND seems to improve dynamics, as well as a very impressive bottom end. It is being used by many members with a modified Silicon Chip Headphone Amplifier design, using kits from Jaycar and Altronics. (Australia)
Many members in Rock Grotto are now reporting fabulous results when it uses 2x 2,200uF low ESR in parallel , where JLH used 2 x ordinary electros.Even better, is the JLH when the current limiter section is bypassed, and the unit connected directly to 78xx/79xx or LM317/LM337 . The initial high switch on surge is handled by the current limiting in the VREGs.We are now using a smaller dual polarity PCB, which due to member interest after impressive posted results, is being supplied to members for little more than cost, by an ex Broadcast engineer, now lecturer.
SandyK
Hi Sandy- I never went further with my preamp, as I think the PS is sufficient using various films and Oscon caps. If I ever get back to my headphone or power amps, I'll certainly try the circuit, as IMO you can't get there with caps alone.
HiFi- In general DF will remain more or less constant over frequency until inductance takes over. Not completely constant, but enough for our purposes. You can calculate esr from DF, and you'll find it varies hugely with frequency (it's a mathematical necessity). That's the reason simulations are useless unless you use a much better cap model (there have been excellent threads here on this). It's easy to get low Z at low frequencies with big electrolytics- X is low and R is low. It's easy to get low Z at very high frequencies because C is reasonably small to produce the needed X, and good film caps excel at DF, so R is low. It's the middle where life gets harder. Let's say you want Z<1 ohm at 3000 Hz. You need at least 53.1uF, so that rules out small film caps. But, there is no electrolytic made with low enough esr and esl to hit 1 ohm at that frequency. An Oscon can get close, but that limits you to low voltages. Thus, depending on what impedance limit you set for yourself, an active circuit may be the only way to get there- or those huge polypropylene motor run caps!
HiFi- In general DF will remain more or less constant over frequency until inductance takes over. Not completely constant, but enough for our purposes. You can calculate esr from DF, and you'll find it varies hugely with frequency (it's a mathematical necessity). That's the reason simulations are useless unless you use a much better cap model (there have been excellent threads here on this). It's easy to get low Z at low frequencies with big electrolytics- X is low and R is low. It's easy to get low Z at very high frequencies because C is reasonably small to produce the needed X, and good film caps excel at DF, so R is low. It's the middle where life gets harder. Let's say you want Z<1 ohm at 3000 Hz. You need at least 53.1uF, so that rules out small film caps. But, there is no electrolytic made with low enough esr and esl to hit 1 ohm at that frequency. An Oscon can get close, but that limits you to low voltages. Thus, depending on what impedance limit you set for yourself, an active circuit may be the only way to get there- or those huge polypropylene motor run caps!
SandyK,
Would you mind pointing to where I can obtain a schematic?
Conrad,
I don't mind using up to 10,000uF for just a couple of opamps! These low ESL capacitors (Panasonic FC and Rubycon ZL) would have a Z 0.008R at 100kHz! I don't know what is the total impedance (resistence and inductive reactance) at lower frequencies such as 1000Hz and 100Hz).
I don't understand how while DF varies little the ESR that can be calculated from DF can vary greatly! I may do a search and find the formula.
I had the luck with putting on 11,000uF behind the LM317/LM337 regulators with LN5819 1A diodes without burning the diodes. But I only tried it once. I have currently 6,900uF behind the regulators and the diodes appear to be safe, as I have already turned the device on and off for dozens of times. They bear the initial surge current for a few seconds without getting burnt. I don't have any current limiting. If needed I would prefer to put on the LN5822 which can stand 3A that costs only a few extra cents, than be bothered with adding additional current limit components. But how big the capacitance that would cause concern?
Another thing I am thinking about is to build something to run in the capacitors. Would it be fine if I just put them after a bridge rectifier with about 23V to run in the 25v caps?
Regards,
Bill
Would you mind pointing to where I can obtain a schematic?
Conrad,
I don't mind using up to 10,000uF for just a couple of opamps! These low ESL capacitors (Panasonic FC and Rubycon ZL) would have a Z 0.008R at 100kHz! I don't know what is the total impedance (resistence and inductive reactance) at lower frequencies such as 1000Hz and 100Hz).
I don't understand how while DF varies little the ESR that can be calculated from DF can vary greatly! I may do a search and find the formula.
I had the luck with putting on 11,000uF behind the LM317/LM337 regulators with LN5819 1A diodes without burning the diodes. But I only tried it once. I have currently 6,900uF behind the regulators and the diodes appear to be safe, as I have already turned the device on and off for dozens of times. They bear the initial surge current for a few seconds without getting burnt. I don't have any current limiting. If needed I would prefer to put on the LN5822 which can stand 3A that costs only a few extra cents, than be bothered with adding additional current limit components. But how big the capacitance that would cause concern?
Another thing I am thinking about is to build something to run in the capacitors. Would it be fine if I just put them after a bridge rectifier with about 23V to run in the 25v caps?
Regards,
Bill
Last weekend, I added 1uF polypropylene as bypass. The tonal balance was better but the sound became harsh possibly because the polypropylene caps were brand-new without any run-in.
I wonder if I increase the resistence from the rail to ground to around 3k would a 20uF MKP bring down the ESR to the mid frequency region?
I wonder if I increase the resistence from the rail to ground to around 3k would a 20uF MKP bring down the ESR to the mid frequency region?
HiFiNutNut
I won't post the original schematic in case of copyright violations.
The attached schematic is what we are currently using. Mainly due to space considerations with a smaller PCB, the 4 x 1,000uF have been replaced by 2 x 2,200uF 10V Low ESR (Jaycar)
As stated previously, only bypass the current limiter stage, when fed directly from LM78xx/LM79xx or LM317/LM337
My apologies for the size of the schematic, I am used to them being auto scaled down !
URL=http://imageshack.us]
I won't post the original schematic in case of copyright violations.
The attached schematic is what we are currently using. Mainly due to space considerations with a smaller PCB, the 4 x 1,000uF have been replaced by 2 x 2,200uF 10V Low ESR (Jaycar)
As stated previously, only bypass the current limiter stage, when fed directly from LM78xx/LM79xx or LM317/LM337
My apologies for the size of the schematic, I am used to them being auto scaled down !
URL=http://imageshack.us]
An externally hosted image should be here but it was not working when we last tested it.
[/URL]
SandyK,
Thanks for your posts.
I don't know if that would solve my specific problem.
I see the issues with using LM317/LM337 as the following:
(1) The LM317/LM337 regulators provide very low impedance at low frequencies.
(2) The bypass capacitors provide low impedance at higher frequencies.
(3) The inductive output of the regs will cause excessive ringing without sufficient ESR (at least 0.5R) for the capacitors. Adding resistance to the bypass capacitors (or relying on their large ESR) is compulsary when using the LM317/337. But these resistence increases the output impedance of the regulators, which causes errors.
(4) Without a large capacitance transient response suffers.
Now the JLH "ripple eater", will it solve all of the above problem?
Can I buy the boards from Jaycar or Altronics?
Regards,
Bill
Thanks for your posts.
I don't know if that would solve my specific problem.
I see the issues with using LM317/LM337 as the following:
(1) The LM317/LM337 regulators provide very low impedance at low frequencies.
(2) The bypass capacitors provide low impedance at higher frequencies.
(3) The inductive output of the regs will cause excessive ringing without sufficient ESR (at least 0.5R) for the capacitors. Adding resistance to the bypass capacitors (or relying on their large ESR) is compulsary when using the LM317/337. But these resistence increases the output impedance of the regulators, which causes errors.
(4) Without a large capacitance transient response suffers.
Now the JLH "ripple eater", will it solve all of the above problem?
Can I buy the boards from Jaycar or Altronics?
Regards,
Bill
Hi HiFi- That big low esr cap is performing the way you think only in your dreams and in the marketing people's special world. A vector impedance meter or HF bridge tells all, and inductance gets you way before 100kHz. Here are a couple things that will help, but IMO nothing is as useful as measurements.
Conrad's Impedance Converter
Handy Electronic Formulas
The converter doesn't have every possible combination, though I do update it now and then. The formula sheet also gets periodic updates, but should have most of the L and C relationships you need.
Regards,
CH
Conrad's Impedance Converter
Handy Electronic Formulas
The converter doesn't have every possible combination, though I do update it now and then. The formula sheet also gets periodic updates, but should have most of the L and C relationships you need.
Regards,
CH
Hi Bill !
The JLH has a very low output impedance to several hundred KHZ. JLH stated .02 ohms to beyond 150KHZ
The use of the JLH results in a very natural presentation because of it's low impedance, and in JLH's own situation, it lowered noise and ripple from his bench PSU from 300uV to 4uV.
It is not a commercial product, and the etching of the small PCB shouldn't be too hard for most DiyAudio members.
I recently completed another JLH, but I haven't tried letting the smoke out yet. You are welcome to borrow it for a week or so and evaluate it. Even post the results if you wish.
I will email you a copy of the original JLH article .
Regards
Alex
The JLH has a very low output impedance to several hundred KHZ. JLH stated .02 ohms to beyond 150KHZ
The use of the JLH results in a very natural presentation because of it's low impedance, and in JLH's own situation, it lowered noise and ripple from his bench PSU from 300uV to 4uV.
It is not a commercial product, and the etching of the small PCB shouldn't be too hard for most DiyAudio members.
I recently completed another JLH, but I haven't tried letting the smoke out yet. You are welcome to borrow it for a week or so and evaluate it. Even post the results if you wish.
I will email you a copy of the original JLH article .
Regards
Alex
Andrew
The voltage regulators limit the brief high initial current inrush.There is a noticeable improvement in audible performance in most cases by bypassing the current limiter. The current limiter is of course necessary with high voltage supplies. There have now been quite a few of these built,and used in this way.( Even in your country.) In fact a prototype headphone amplifier( see photo) using this method was sent to Scotland for evaluation. I have been using a JLH with 7815/7915 regulators in this fashion for some considerable time. There have been no reported problems. If you would like a copy of the original article from ETI, send me an email and I will forward it.
SandyK
The voltage regulators limit the brief high initial current inrush.There is a noticeable improvement in audible performance in most cases by bypassing the current limiter. The current limiter is of course necessary with high voltage supplies. There have now been quite a few of these built,and used in this way.( Even in your country.) In fact a prototype headphone amplifier( see photo) using this method was sent to Scotland for evaluation. I have been using a JLH with 7815/7915 regulators in this fashion for some considerable time. There have been no reported problems. If you would like a copy of the original article from ETI, send me an email and I will forward it.
SandyK
I have swapped in and out various caps and am standing by my original assessments.
Each time when I swapped in the Rubycon ZL I liked the sound (most neutral) but after a while could not live with the peaky treble. The tweeter was padded with more resistance and even a 50uH inductor. But the sound was still peaky. Each time when I swapped in the Panasonic FC I was happy because the peaky treble was gone, however, after a while I could not live with the slight metallic coloration.
I also tried a pair of the old blue coat Phillips (without running in). The sound was very smooth but appeared to be very slow and a complete lack of details. I may build a device to run in these caps.
This points me to searching for a regulator that can eliminate dependency on caps. But are there any regulators that provide very good transient response without relying some large caps on the output? or at least, a regulator that can allow LOW ESR/ESL caps without going into oscillation? Jung regulator? Borbely?
Note that if I remove the resistance after the LM317/337 and let the things ring, although in the resonant frequency the sound is messy, it does to a great extent remove the colorations of the caps.
Each time when I swapped in the Rubycon ZL I liked the sound (most neutral) but after a while could not live with the peaky treble. The tweeter was padded with more resistance and even a 50uH inductor. But the sound was still peaky. Each time when I swapped in the Panasonic FC I was happy because the peaky treble was gone, however, after a while I could not live with the slight metallic coloration.
I also tried a pair of the old blue coat Phillips (without running in). The sound was very smooth but appeared to be very slow and a complete lack of details. I may build a device to run in these caps.
This points me to searching for a regulator that can eliminate dependency on caps. But are there any regulators that provide very good transient response without relying some large caps on the output? or at least, a regulator that can allow LOW ESR/ESL caps without going into oscillation? Jung regulator? Borbely?
Note that if I remove the resistance after the LM317/337 and let the things ring, although in the resonant frequency the sound is messy, it does to a great extent remove the colorations of the caps.
HiFi- since I'm always interested in testing components looking for solid reasons for what people hear, what common capacitor would you say is worst at the adjustment pin? Or any other location? Something really stinky bad that even I might be able to hear, but not a known offender like tantalum. Just a commonly available aluminum electrolytic.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.