Yeah. Anyone who works with tube circuits or even AC line voltages has occasionally gotten zapped, or at least a tickle. Every time its happened to me, its been a strong reminder about safety. First time I can remember getting a shock was as a teenager when I was hooking up an extension phone for my bedroom. Just as I was stripping the final wires working flat on my belly in the crawlspace under the house, the phone rang. The ringer voltage is about 100VAC which I didn't know was that high until the very moment it rang.
You know, anyone who says they haven't got zapped a few times when they knew better is either not telling the truth, or is so careful they aren't getting anything done. Electricians working with 550 / 600 can get killed the first time out easily, and they are never quite the same again. So where there is a high chance of death or disfigurement, my earlier comment doesn't apply. Alive and breathing counts.
Early, working on tube TVs, you could get nailed without even touching the 2nd anode (I found out and saw the spark jump the gap). I've touched the focus control behind the front panel of an oscilloscope a couple times (different instruments). Those things get your attention in a way that creates a lasting memory! lol! Not once have I been zapped in a lab environment.
Every time you consider how stupid you are. Then there are voltages that can show up where they aren't supposed to.
Hey Mark, ringer is close (90 VAC spec), but a DSL signal will absolutely get your attention. It's fun when the telco brings up the DSL on the wrong pair - unexpected.
Hi Hierfi,
Try to keep in mind that personal attacks are generally not the intent for most of us, myself included. "Attack the idea, not the person"
Early, working on tube TVs, you could get nailed without even touching the 2nd anode (I found out and saw the spark jump the gap). I've touched the focus control behind the front panel of an oscilloscope a couple times (different instruments). Those things get your attention in a way that creates a lasting memory! lol! Not once have I been zapped in a lab environment.
Every time you consider how stupid you are. Then there are voltages that can show up where they aren't supposed to.
Hey Mark, ringer is close (90 VAC spec), but a DSL signal will absolutely get your attention. It's fun when the telco brings up the DSL on the wrong pair - unexpected.
Hi Hierfi,
No warning there. It's instant.
Try to keep in mind that personal attacks are generally not the intent for most of us, myself included. "Attack the idea, not the person"
Last edited:
"No warning there. It's instant."
Thats what I thought... that's why we hired a pre-cog highly recommended to us by Peter Venkman. Apparently he switched his research from ESP based paranormal phenomena to that of Pre-Cog. Never could tell if she was worth the money though. Always seemed to come with some excuse when someone got zapped.
I'm not sure if I would think to attack the idea, rather to question if the conclusions follow from the premises. To attack the idea can be perceived as an attack on those holding the ideas. If I do engage in responses it is likely only a defensive response.
Thats what I thought... that's why we hired a pre-cog highly recommended to us by Peter Venkman. Apparently he switched his research from ESP based paranormal phenomena to that of Pre-Cog. Never could tell if she was worth the money though. Always seemed to come with some excuse when someone got zapped.
I'm not sure if I would think to attack the idea, rather to question if the conclusions follow from the premises. To attack the idea can be perceived as an attack on those holding the ideas. If I do engage in responses it is likely only a defensive response.
Last edited:
Well, there you go!
Pre-crime (good movie featuring this idea), or pre-anything. Wouldn't that be wonderful?
Anyway, let's get back on topic please.
Pre-crime (good movie featuring this idea), or pre-anything. Wouldn't that be wonderful?
Anyway, let's get back on topic please.
I've been looking into the problem of increased distortion in aluminium electrolytics (AE's) trying to understand how this happens. We know it is not because of the dielectric because we can model it quite accurately, but we do not have a model for the electrolyte part of an AE at this point. However, some preliminary scouting around on the web seems to indicate that electrolyte conductivity IS frequency dependent iao things - so it is certainly not a perfect conduction medium. So the question is, is this the smoking gun? Electrolyte conductivity is exclusively ionic, whereas in a conductor like copper, it is electron-based and it is not as efficient as electron conduction. I'll keep digging to see what I can find. Maybe someone on this thread has more info on this.
Just for reference, the aluminium oxide electron cloud deformation (the mechanism that enables charge to be stored) itself used as the dielectric in AE's has an upper-frequency limit of 10^17 Hz and it stores charge quite happily for weeks or months - in other words, it is also quite happy at DC as well. This
Just for reference, the aluminium oxide electron cloud deformation (the mechanism that enables charge to be stored) itself used as the dielectric in AE's has an upper-frequency limit of 10^17 Hz and it stores charge quite happily for weeks or months - in other words, it is also quite happy at DC as well. This
Last edited:
Excellent that your looking Bonsai. Never be sure something else isn't going on. You can't simulate what you don't know exists.
I've engaged in lots of variant research initiates providing electrical and electronic assistance getting into the electron levels with currents in 100's of electrons/sec, including providing assistance in the modelling of biological materials for the purpose of replacing conventional electrical memory with biological memory. Thought it was really farfetched, though someone wrote a book about it. The resistance gets so high that noise becomes overwhelming and you get into using charge coupled devices for your signal path.
I've engaged in lots of variant research initiates providing electrical and electronic assistance getting into the electron levels with currents in 100's of electrons/sec, including providing assistance in the modelling of biological materials for the purpose of replacing conventional electrical memory with biological memory. Thought it was really farfetched, though someone wrote a book about it. The resistance gets so high that noise becomes overwhelming and you get into using charge coupled devices for your signal path.
How an electrolytic capacitor is constructed and operates should be common knowledge for technicians. Basic stuff. I agree, you need to understand the parts you are using in order to apply them properly. Same for mechanics or any other professional.
And there is the danger of simulators. Often they don't get all the variables accounted for and end up with erroneous results. You should see the new circuit simulators Keysight is using. You need a degree to use them though, it isn't just plugging in numbers. Simply testing things empirically at least gives you truth if you run the test under expected circuit conditions.
Electron flow at those low levels will be dominated by noise if you aren't very careful. Anything dealing with biology becomes very complicated, very quickly and time is a definite concern. I've looked into it a tiny bit compared to what you have.
And there is the danger of simulators. Often they don't get all the variables accounted for and end up with erroneous results. You should see the new circuit simulators Keysight is using. You need a degree to use them though, it isn't just plugging in numbers. Simply testing things empirically at least gives you truth if you run the test under expected circuit conditions.
Electron flow at those low levels will be dominated by noise if you aren't very careful. Anything dealing with biology becomes very complicated, very quickly and time is a definite concern. I've looked into it a tiny bit compared to what you have.
Some unflattering posts were removed.
Please stick to facts, in accordance with OP's well elaborated introduction.
Hugo
Moderator team.
Please stick to facts, in accordance with OP's well elaborated introduction.
HugoModerator team.
Not dissing electronics techs here, but I doubt the average electronics tech would know much about how wet chemistry behaves electrically. I've only scratched the surface on the topic because I developed a conductivity imaging system for my master's thesis. The system sensed the conductivity in an array of 'voxels' (volume pixels). This was intended for counting cancer cells in a sample. Life is fun when the electrodes corrode at the surface molecular level and you have to design a circuit that works around that.
I'm not an expert on how aluminum electrolytic capacitors work on the inside, but like many I know the basics: Etched film to get maximum surface area with some sort of soup contained in a medium to drive up the capacitance. How that behaves as function of frequency could get tricky as @Bonsai hints at. I'm curious to see Part II of that story. 🙂
Well, no. You end up with imprecise results. Not necessarily erroneous. If you're building an RC circuit to set the LF pole at 2 Hz and capacitors are only available following the E6 standard values at ±20 % tolerance, is it honestly important to account for some secondary or tertiary effect? I would argue that anyone who specifies that cutoff frequency with more than two – or maybe even one – significant digit is missing the point. 1/2pi*RC gets you close enough to reality.
But if you're trying to design an amp that needs -140 dBc THD around that cutoff point the simple capacitor model is not always enough.
There's actually nothing inherently wrong with SPICE or Spectre simulators. They both support pretty fancy modelling. The issue is mainly in the device models, at least for the devices that those of us who are on the outside of the device design houses have access to. These models are often simplified. In part because the average consumer of those devices wouldn't benefit from the added precision but also to ensure that the simulations will even converge to a DC operating point. I've lost years of my life to convergence issues in computer simulators. The more complexity you throw into the model the harder it becomes for the computer to find a solution to the equation system.
The device models we had when I was at National Semiconductor and TI were remarkably accurate. Even for simulations of complex phenomena like phase noise in PLLs the simulation results were very, very close to reality. And that was at a performance level similar to what -140 dBc THD would be in audio. It helps a lot when you have an entire team of employees who's only job is to make models for the devices that you're using.
There is just no way we would have high-end electronic devices today without computer simulation.
Well.... It gives you a truth. 🙂 I'm with you to a point, though. There's nothing quite like powering up a new circuit and comparing its performance against the simulations. But it only shows the performance of one part. You need to test thousands to get an idea of part-to-part variation. And it would be enormously impractical to design circuits by changing components in the lab. It's possible, of course. Some do that. But changing components is a hassle and changing to a component that has a different temperature coefficient may require a few days for a new part to get in (versus a single click in a simulator).
So don't throw the simulators out with the bathwater. I agree with your fundamental premise that garbage in equals garbage out, though.
Tom
"Not dissing electronics techs here, but I doubt the average electronics tech would know much about how wet chemistry behaves electrically."
Tom... why would you start out with this statement? Of course you are. The "average" PHD or physicist wouldn't know that either. Are you saying all "average" masters thesis candidates know a lot about how wet chemistry behaves electrically? That is clearly not what anatech intended to message.
Notwithstanding the quality of the substance that followed what I see is that you are propping yourself up to add more weight to substance than it otherwise might deserve, and dissing electronics techs to do it. Otherwise, why make the statement at all?
Tom... why would you start out with this statement? Of course you are. The "average" PHD or physicist wouldn't know that either. Are you saying all "average" masters thesis candidates know a lot about how wet chemistry behaves electrically? That is clearly not what anatech intended to message.
Notwithstanding the quality of the substance that followed what I see is that you are propping yourself up to add more weight to substance than it otherwise might deserve, and dissing electronics techs to do it. Otherwise, why make the statement at all?
The problem is even at 1-2 Hz, the displacement current electron flow is still high in physical terms, so it's unlikely that it is the cause. Cyril Batemans's excellent articles on capacitor sound don't seem to offer a cogent explanation for why AE's show a rise in distortion at LF as a voltage develops across the cap terminals, although he did show the differences in performance of the various dielectrics.
How a cap works at the atomic level within a dielectric is not basic stuff. You have to invoke Maxwell's electric field equations to understand how the electron clouds with the atoms are deformed. You wont need to know this stuff to understand how to use a cap (I certainly don't), but it does help explain why some dielectrics are very superior NP0/COG for example) and others not so (X7R, but even that under certain conditions is very low distortion if I understand Bateman's measurements correctly).
As soon as you dig into the detail of these things, it quickly becomes a very specialist field, and someone not involved in that field intimately probably will not be aware of the complexities involved.
Hi Tom,
The models you referenced are very good. The ones Keysight and others use also model the exact PCB, materials, vias and other things that affect performance at HF. The people who run those have that as their entire job.
I'm talking about the simple simulators we commonly see people use here on this site and others on the same level. They do not include the rather more important variables that you did at TI for example. You are quite right though, you can't test a population of completed circuits economically before deployment (but you better check a few!).
My message is really that at the levels you commonly see here, you cannot trust the results entirely. Useful - yes. The last word, heck no! Not even close.
Relating to capacitors, we have models, we know tends (like leakage vs temperature and voltage) and stuff like that. But for timing or filter use, electrolytic capacitors are only useful for very approximate applications. Absolutely correct.
Also, your comment on accuracy really applies to many things. With a digital meter, people will quote the entire reading as the proper value without regard to an error budget or possible circuit loading, noise or crest factor (for AC readings), or even calibration best case. So we have a basic lack of understanding on almost everything.
The models you referenced are very good. The ones Keysight and others use also model the exact PCB, materials, vias and other things that affect performance at HF. The people who run those have that as their entire job.
I'm talking about the simple simulators we commonly see people use here on this site and others on the same level. They do not include the rather more important variables that you did at TI for example. You are quite right though, you can't test a population of completed circuits economically before deployment (but you better check a few!).
My message is really that at the levels you commonly see here, you cannot trust the results entirely. Useful - yes. The last word, heck no! Not even close.
Relating to capacitors, we have models, we know tends (like leakage vs temperature and voltage) and stuff like that. But for timing or filter use, electrolytic capacitors are only useful for very approximate applications. Absolutely correct.
Also, your comment on accuracy really applies to many things. With a digital meter, people will quote the entire reading as the proper value without regard to an error budget or possible circuit loading, noise or crest factor (for AC readings), or even calibration best case. So we have a basic lack of understanding on almost everything.
Sticking my nose in here and wholeheartedly agreeing with @anatech that LTSpice and other hobbyist type simulations will only get you close - and that's even if you do what almost no one ever bothers - or knows - to do and include parasitics and such in the simulation. Even the semiconductors, especially the IC's, are typically at best macro models - which makes sense as the manufacturers have very valid IP in the products.
With all that being true, it's really quite amusing to see some threads go on for literally hundreds of posts arguing with religious zeal the merits of someone's pet circuit and the effects of changing some minor component in an insignificant way... and not a single person has ever actually bothered to build the damn thing to see how it really operates. Sigh.
Hal
With all that being true, it's really quite amusing to see some threads go on for literally hundreds of posts arguing with religious zeal the merits of someone's pet circuit and the effects of changing some minor component in an insignificant way... and not a single person has ever actually bothered to build the damn thing to see how it really operates. Sigh.
Hal
And the same is true for those same lengthy posts attacking a design by someone who has spent a lifetime successfully designing said circuits, all because they throw together an incomplete simulation, get different results, and automatically think that they not only can do better but have somehow achieved world peace by doing so - and throw a tantrum if anyone who actually knows what they are doing dares to call BS.
I think this thread got started because a claim was made by a manufacturer that Capacitors don't distort or don't distort that much. Bateman's practical experience and measurements say that some capacitors do. We then had a statement made that DA causes sound problems, and this has been discussed on diyAudio before, and the conclusion was that DA in itself does not cause distortion, and a model for DA behaviour was offered up as a proof point to show that DA is shifted down in frequency to where it really is not an issue i.e. information is not getting released from the DA after the stimulus that caused it that will show up as harmonic distortion.
It seems wrt AE's, there is a lot not known about wet electrolytes and it could be this is where the problem lies with AEs and not in the aluminium oxide dielectrics. Rooting around on the web, it seems that the conductance of some electrolytes is not linear with frequency.
It seems wrt AE's, there is a lot not known about wet electrolytes and it could be this is where the problem lies with AEs and not in the aluminium oxide dielectrics. Rooting around on the web, it seems that the conductance of some electrolytes is not linear with frequency.
With all due respect, I did not interpret the model exhibits to show that DA effects are always shifted down in frequency so as to be inaudible. The test signals that showed the most dramatic effects were included in a published article. However, I am told by one of the authors that the article and measurements resulted from trying to find a way to show that DA can alter waveforms dependent on the particular waveform. For some audio waveforms the effects can be audible, which is what prompted the authors to look for a way to measure some effect, any effect at all, to prove the point that DA can alter waveforms in a waveform dependent way.
Talking about simulators, don’t forget that there are some very powerful IC process simulators out there as well. Back when I was a PL GM nearly 20 yrs ago, our discrete device process dev guys were using t-cad. Things have probably on from there, I don’t know.
Hi Mark,
Any testing has to put the part in the same circumstances it sees in audio circuits and at the frequencies we use. It's easy to show a component misbehaving, especially when used outside the normal use. So looking at data where the part isn't used in your application is a complete waste of time, and is in fact misleading.
Any testing has to put the part in the same circumstances it sees in audio circuits and at the frequencies we use. It's easy to show a component misbehaving, especially when used outside the normal use. So looking at data where the part isn't used in your application is a complete waste of time, and is in fact misleading.
I'd love to have an EM field solver (and person to run it) for my work. All I can do without spending thousands on field solver software that I wouldn't have time to learn is to visualize current densities in my head. It sure would be nice to be able to check my hypotheses.
True that. LTspice and others also assume that a BC547B is a BC547B never mind that a BC547B may be subtly (but quantifiably) different from manufacturer to manufacturer. Then add behavioural models like the LM3886 and many (but not all) bets are off.
All models are wrong but some are useful. 🙂
Tom
Simulate what you build, build what you simulate. Exactly. Then you can argue about why the two are different.
Tom, Tesla and Maxwell had great EM field visualization ability as well. Maxwell thought in terms of circulating currents (the vector Potential) and Tesla visualized the fields, currents and mechanics in motors and generators running. The simulators were all in their heads. Good company! 👍
Tom, Tesla and Maxwell had great EM field visualization ability as well. Maxwell thought in terms of circulating currents (the vector Potential) and Tesla visualized the fields, currents and mechanics in motors and generators running. The simulators were all in their heads. Good company! 👍