Fanuc said:
Kevin,
That's why in the original 1994 series we recommended remote sensing, if you wanted to duplicate the test bench performance numbers in a real application. In contrast to what most people seem to think, we really DID think about it 😉
Jan Didden
Bob Cordell said:
Hello Bob,
Sorry for the delay replying to this post. Have been very busy at work and at the house etc. I guess you are right about this. I think we could say that the Zout would be reduced by 20db more realistically with an iLoad of 50mA. The PSRR would be degraded by some margin at all frequencys also but I am not sure by what.
What I mean't was that with a very low Zout on the datasheet spec'd at 500mA (unrealistic) plus the PCB resistances this would reduce the Zout figures you mention considerably.
I have read in numerous datasheets that using zeners like 20v or whatever to tee off the LM317/LM337 to get 50+ volts reduces the performance of the reg. As I said above by what I margin I don't know. Having said that I am aware of the LM117HV or LM317HV types that might be better here.
This raises some good questions about the perfomance of the capacitors used to do this at High Voltage. RC filtering seems very effective in digital designs at low voltage - the caps available are excellent. OSCONs, polymer types and MLCC ceramics etc.
Best Regards
Kevin
Fanuc said:
Hi Kevin,
I agree. Never overlook what a simple RC can do for you in terms of PSRR and noise suppression at high frequencies, but at the same time realize that they are not perfect, and have real ESR and ESL.
BTW, one thing I learned the hard way a couple of years ago on a real project on my day job was how some of the low dropout regulators detest seeing a low-ESR capacitor right at their output. A case where seemingly better is not better. They oscillated and misbehaved in all sorts of wierd ways. So I just put a 100 uF tantalum directly at the output to ground, then went through a small 1 uH inductor, and only then hung on the low-ESR capacitors.
Cheers,
Bob
SPICEing Power Supplies
Has anyone here ever used SPICE to help answer some power supply questions, such as to look at rectifier diode peak currents, the use of rectifier snubbers, the influence of parasitic inductances, etc in power supplies?
Cheers,
Bob
Has anyone here ever used SPICE to help answer some power supply questions, such as to look at rectifier diode peak currents, the use of rectifier snubbers, the influence of parasitic inductances, etc in power supplies?
Cheers,
Bob
Re: SPICEing Power Supplies
Hi Bob,
Tom Gootee has done a lot of work in this area, and I've done a little bit. Check out this thread in the Power Supply Design forum.
Bob Cordell said:
Hi Bob,
Tom Gootee has done a lot of work in this area, and I've done a little bit. Check out this thread in the Power Supply Design forum.
Hi Andy,
Thanks for the mention. But I think that you are being too modest. Your (and powerbecker's) contributions to the spice transformer-modeling work that I/we did were quite significant! I am still just a novice, at most of this stuff.
Bob, et al,
Our resulting transformer model's schematic and downloadable LTspice simulation file are at:
http://www.fullnet.com/~tomg/gooteesp.htm
At that same URL, there is also a schematic and downloadable LTspice files for a linear power supply, which uses our spice model to model a(nother) real transformer.
That spice transformer model is pretty handy, because it uses very-easily-obtained measurements, and then automatically calculates the rest of the required parameters for the model. And the procedures for making the measurements are included, right on the schematic.
I have used that transformer model to help successfully design rectifier snubbers, soft-start circuits, etc, and have also used it to aid in the design of the off-line boost-mode switching power supply for my Curve Tracer product, all with LTspice.
The model is based on the method outlined on page 4 of the following paper:
http://www.onsemi.com/pub/Collateral/AN1679-D.PDF
There is another thread currently underway/resurrected, on modeling transformers with spice, at:
http://www.diyaudio.com/forums/showthread.php?s=&postid=1283540#post1283540
Another paper about modeling transformers was mentioned, in that thread:
http://www.eettaiwan.com/ARTICLES/2...DA_DA_AN215.PDF
Regards,
Tom Gootee
http://www.fullnet.com/~tomg/index.html
Thanks for the mention. But I think that you are being too modest. Your (and powerbecker's) contributions to the spice transformer-modeling work that I/we did were quite significant! I am still just a novice, at most of this stuff.
Bob, et al,
Our resulting transformer model's schematic and downloadable LTspice simulation file are at:
http://www.fullnet.com/~tomg/gooteesp.htm
At that same URL, there is also a schematic and downloadable LTspice files for a linear power supply, which uses our spice model to model a(nother) real transformer.
That spice transformer model is pretty handy, because it uses very-easily-obtained measurements, and then automatically calculates the rest of the required parameters for the model. And the procedures for making the measurements are included, right on the schematic.
I have used that transformer model to help successfully design rectifier snubbers, soft-start circuits, etc, and have also used it to aid in the design of the off-line boost-mode switching power supply for my Curve Tracer product, all with LTspice.
The model is based on the method outlined on page 4 of the following paper:
http://www.onsemi.com/pub/Collateral/AN1679-D.PDF
There is another thread currently underway/resurrected, on modeling transformers with spice, at:
http://www.diyaudio.com/forums/showthread.php?s=&postid=1283540#post1283540
Another paper about modeling transformers was mentioned, in that thread:
http://www.eettaiwan.com/ARTICLES/2...DA_DA_AN215.PDF
Regards,
Tom Gootee
http://www.fullnet.com/~tomg/index.html
gootee said:
Tom, since your name came up in this context, I'd love to know if you've ever modeled a Jensen 4 pole electrolytic. I've posted in Electronic Parts a couple of times in the past about this, and never got an answer. I even emailed Jensen to ask if they has a Spice model for this design. I had to explain to them what a Spice model was! Never heard back from them.
tonyptony said:
I haven't. Sorry.
I remember hearing about them, but was never even clear about what they were, actually. You might want to search and then maybe ask in the sci.electronics.design group, at http://groups.google.com , or one of the rec.audio.* groups.
But I could SWEAR that I saw, within the last month or so, some four-lead capacitors, from some other manufacturer, listed on (I think) the mouser.com website. But I tried searching at mouser, just now, and couldn't find anything like that. Maybe it was digikey.com. And the only other one I remember even looking at, that recently, was alliedelec.com.
At any rate, if they WERE something similar, and you can find the listing, there's a good chance that they might have a datasheet, plus more info on the manufacturer's website. And if you're REALLY lucky, maybe it would be one of the manufacturers who has lots of spice models (e.g. Kemet).
Of course, if it's the Jensens that you want, specifically, then none of that would help you, maybe.
But if you HAVE one of the caps that you want to model, then you "should" be able to do that from measurements. There are technical papers, theses, appnotes, etc, on the web, about doing just that. But you might then wish that you had a nice HP/Agilent Network Analyzer, or something similar.
There are some very good and very sophisticated techniques for capacitor SPICE modeling, with frequency-varying parasitics, higher-order effects, etc, etc, etc, to be found in the Files section of the LT-SPICE group, at http://www.yahoogroups.com , plus many message-threads about capacitor modeling in the message archive, there, and, last but not least, some brilliant, helpful people who might answer, if you post a question.
Good luck! Sorry I couldn't be of more help.
P.S. Your story about having to explain to someone at Jensen about what a Spice model is made me smile. I had exactly the same experience with someone from a large thermistor manufacturer, not too long ago. I then got an email from some higher-up, there, who seemed very interested but had no clue, whatsoever, about Spice, or modeling, etc. I wrote him a page-long reply, explaining, quite well I think, the prevalence and usefulness of Spice, etc. And I pointed out to them that Epcos (one of their biggest competitors) had a VERY nice spice-model library, with all of their hundreds of thermistors, with great ready-to-run models that even included self-heating effects, etc, and described a few simple examples of the major benefits I had derived from using it. And I also opined to them that if THEY had a library of spice models for THEIR components, that they might very well SELL a lot more of them! I have not heard back from them. But at least they were interested.
- Tom Gootee
http://www.fullnet.com/~tomg/index.html
Tom, Sprague used to make 4 lead electrolytic caps for switching supply operation. I have a bunch of low voltage units at my warehouse. They seemed to go out of style over the years.
Try patent # 4509100 from Sprague. (Brought to you by a Google search in less than 5 minutes)
Try patent # 4509100 from Sprague. (Brought to you by a Google search in less than 5 minutes)
john curl said:Tom, Sprague used to make 4 lead electrolytic caps for switching supply operation. I have a bunch of low voltage units at my warehouse. They seemed to go out of style over the years.
Try patent # 4509100 from Sprague. (Brought to you by a Google search in less than 5 minutes)
Thank you, John.
john curl said:Tom, Sprague used to make 4 lead electrolytic caps for switching supply operation. I have a bunch of low voltage units at my warehouse. They seemed to go out of style over the years.
Try patent # 4509100 from Sprague. (Brought to you by a Google search in less than 5 minutes)
John, Tom,
BHC Aerovox are also making what they call slit foil electros.
I believe it is the same concept as Jensen.
http://www.bhc.co.uk/slit_foil.htm
I haven't done any measurements on them but the 4 pole
Jensens are very good capacitors purely based on listening tests.
cheers
Terry
Hi all.
Thanks to Bob and all of you for a very interesting discussion.
Bob,
You started this tread by saying: I like to use two caps per rail, separate them with a small resistor (typically 0R22.....
This sounds like a good idea.
How about also separating "sentertap/0v capset1" and "0v capset2" with a 0R22. This is to separate the noisy charging ground of capset1 a bit from the not so noisy central reference ground of capset2.
This was briefly mentioned in post 476, but did not encourage any further discussion. Was that so because it is a bad idea? Could you please explain why this is, or is not a bad idea.
Rebone
Thanks to Bob and all of you for a very interesting discussion.
Bob,
You started this tread by saying: I like to use two caps per rail, separate them with a small resistor (typically 0R22.....
This sounds like a good idea.
How about also separating "sentertap/0v capset1" and "0v capset2" with a 0R22. This is to separate the noisy charging ground of capset1 a bit from the not so noisy central reference ground of capset2.
This was briefly mentioned in post 476, but did not encourage any further discussion. Was that so because it is a bad idea? Could you please explain why this is, or is not a bad idea.
Rebone
rebone said:
Let's say you make the 0V_capset2 your Star ground reference. Then you connect a spoke of the Star to 0V_capset1. All of that noise of capset1 will be forced to merge and travel down that spoke. It is on that spoke where you are asking whether or not the addition of a 0.22 ohm resistor can help.
I would tend to say that it would not make much difference, other than to introduce some additional supply drop under current imbalance conditions between the negative and positive rails. The reason I think it might not make much difference is that the 0R22 ohm resistors on the rails between the capacitors of set1 and set2 already provide an impedance to current flow that is much larger than typical copper ground impedances.
Cheers,
Bob
Dual mono designs
There hasn't been much power supply activity here lately, so I thought I'd pose a Devil's Advocate question.
Let's say we are designing a stereo amplifier. We have the choice of a dual mono design using two completely separate power supplies, or we have the choice of using a single power supply.
Most people would instinctively say dual mono, especially based on the way many higher-quality amplifiers have been designed for some time now.
We know dual mono costs more to build and takes up more space, but is it the best way to spend out money and space? Might there be a better way?
Others have made the argument in the past that for the same cost and space, a single supply of larger total VA can be put in place, for example. Then, either channel has full access to all of the energy storage on a transient. Clearly, two 500VA transformers cost more and take up more space than one 1000 VA transformer.
So let's examine the presumed advantages of dual mono design, for which we are paying the extra price in cost and space. Are they real? If so, are there other approaches that could claim the same advantages?
For example, is the main advantage of dual mono design mainly crosstalk? If so, in what part of the frequency band?
If we employ a single supply and quasi-regulate each channel with a capacitance multiplier of some kind, do we then end up with a better solution? The assumption is that the regulator costs little, dissipates little, drops relatively little voltage, and its output generally tracks the average available supply voltage. These may seem like big IFs to some, but let's assume for now it can be done and the various costs of the two "soft rail regulators" can be quantified. Yes, of course, it does add somewhat to the electronic complexity of the amplifier.
Any thoughts?
Lets especially examine with a critical eye what the real advantages of a dual mono design are. Can they be measured? Or do they fall into the unmeasurable category?
Cheers,
Bob
There hasn't been much power supply activity here lately, so I thought I'd pose a Devil's Advocate question.
Let's say we are designing a stereo amplifier. We have the choice of a dual mono design using two completely separate power supplies, or we have the choice of using a single power supply.
Most people would instinctively say dual mono, especially based on the way many higher-quality amplifiers have been designed for some time now.
We know dual mono costs more to build and takes up more space, but is it the best way to spend out money and space? Might there be a better way?
Others have made the argument in the past that for the same cost and space, a single supply of larger total VA can be put in place, for example. Then, either channel has full access to all of the energy storage on a transient. Clearly, two 500VA transformers cost more and take up more space than one 1000 VA transformer.
So let's examine the presumed advantages of dual mono design, for which we are paying the extra price in cost and space. Are they real? If so, are there other approaches that could claim the same advantages?
For example, is the main advantage of dual mono design mainly crosstalk? If so, in what part of the frequency band?
If we employ a single supply and quasi-regulate each channel with a capacitance multiplier of some kind, do we then end up with a better solution? The assumption is that the regulator costs little, dissipates little, drops relatively little voltage, and its output generally tracks the average available supply voltage. These may seem like big IFs to some, but let's assume for now it can be done and the various costs of the two "soft rail regulators" can be quantified. Yes, of course, it does add somewhat to the electronic complexity of the amplifier.
Any thoughts?
Lets especially examine with a critical eye what the real advantages of a dual mono design are. Can they be measured? Or do they fall into the unmeasurable category?
Cheers,
Bob
Hi Bob,
I have to say reliability is the only relative advantage in my opinion.
for high power levels it might be important.
Everything else wil fall on that traditional discussion about sonic enhancement etc.
best regards,
I have to say reliability is the only relative advantage in my opinion.
for high power levels it might be important.
Everything else wil fall on that traditional discussion about sonic enhancement etc.
best regards,
blmn said:
What good is it if one of the two supplies in a dual mono stereo amplifier fails? Not sure that's a strong argument for reliability, but maybe you had something else in mind.
I really do want to hear what people think on this subject. Maybe there are some conditons under which it matters more than others, for example maybe depending on the quality of the PSRR of the amplifier design. Or maybe two separate simple ones are better than one shared simple one, but then maybe a shared single one with some inexpensive per-channel regulation might be better.
Cheers,
Bob
A possible middle alternative is to use one power supply and reverse the input and output leads on one channel. Since bass transients take the most power, and arguably are about equal in both channels at the same time, reversing the polarity of one channel can utilize opposite polarities of the power supply instead of the same polarity at the same time, and not drain them to the same extent.
pooge said:
This is a good point, and actually one of the things I like about balanced (bridged) amplifiers.
Cheers,
Bob
Bob Cordell said:
Yes, it is one reason.
For high power systems is better to have half of the total power than nothing in many cases.
But, as I said, in all other case a good design regarding PSSR and a good choice of components of the PS must be enough for almost all cases.
I built unregulated, regulated and separated PS and, for normal levels of power demands, I can´t see big diferences among them.
best regards,
- Status
- Not open for further replies.