Just curious - anyone have an opinion on Epox boards? I have two K6-2 boards; one Tyan and one Epox. In a lot of ways I like the Expox board better, but it uses Tayeh capacitors, and is more finicky about what power supply it will work with. Like I said, just curious..
If you don't like the caps you can always replace them provided they are the same voltage. 😉
On a far away planet reading your posts in the year 10,191.
There's quite a bit more to it than just getting the same voltage rating. In fact, you don't even necessarily need to use the same voltage rating, and in some cases you can also change the capacitance. But you can't just stick any random capacitor onto a motherboard and expect it to work, even if it's the same value as the original.
As for Epox, they're decent enough boards, but I never really understood what the big deal was... some people swear by 'em, but to me they always seemed completely average.
wrote:There's quite a bit more to it than just getting the same voltage rating. In fact, you don't even necessarily need to use the same voltage rating, and in some cases you can also change the capacitance. But you can't just stick any random capacitor onto a motherboard and expect it to work, even if it's the same value as the original.
As for Epox, they're decent enough boards, but I never really understood what the big deal was... some people swear by 'em, but to me they always seemed completely average.
I've always stayed with the same voltage and never had any issues. Without knowing the circuit in all detail using different voltage caps is risky and having data on everything is just dreaming considering if it is working being done on something ancient.
On a far away planet reading your posts in the year 10,191.
Using a high voltage cap will be fine, make sure you keep the capacitance value the same (f).
Without knowing the circuit in all detail using different voltage caps is risky and having data on everything is just dreaming considering if it is working being done on something ancient.
This is the last I'll say on the matter, since I don't want to derail OP's thread any further, and I really don't feel like trying to teach you basic electronics anyway... but:
It is always safe to raise the voltage rating of a capacitor, and it takes nothing more than a voltmeter and a bit of critical thinking to determine if it's safe to lower it. When in doubt, stay the same or higher, but if you know what you're doing, you can sometimes go lower. No further circuit details needed.
If there is some reason to delve further into it, the circuit details are pretty easy to figure out, since all motherboard VRMs are one of two basic designs. Having the relevant data is not 'dreaming,' it's knowing a bit of theory and maybe looking up a couple datasheets.
But the main point I was getting at is that you didn't even mention ESR, even though that's the single most important detail for caps on modern motherboards. Instead you're getting hung up on the minor shit, which is usually flexible to a certain extent.
wrote:This is the last I'll say on the matter, since I don't want to derail OP's thread any further, and I really don't feel like tryi […]
Without knowing the circuit in all detail using different voltage caps is risky and having data on everything is just dreaming considering if it is working being done on something ancient.
This is the last I'll say on the matter, since I don't want to derail OP's thread any further, and I really don't feel like trying to teach you basic electronics anyway... but:
It is always safe to raise the voltage rating of a capacitor, and it takes nothing more than a voltmeter and a bit of critical thinking to determine if it's safe to lower it. When in doubt, stay the same or higher, but if you know what you're doing, you can sometimes go lower. No further circuit details needed.
If there is some reason to delve further into it, the circuit details are pretty easy to figure out, since all motherboard VRMs are one of two basic designs. Having the relevant data is not 'dreaming,' it's knowing a bit of theory and maybe looking up a couple datasheets.
But the main point I was getting at is that you didn't even mention ESR, even though that's the single most important detail for caps on modern motherboards. Instead you're getting hung up on the minor shit, which is usually flexible to a certain extent.
Save the overinflated ego for other things and no I didn't mention ESR. I have absolutely nothing more to say to you feeling that all you want is a flame war.
On a far away planet reading your posts in the year 10,191.
Flame war? well some of what he said is true a go to bit of can be found on line and mainly data sheets, i glanced over some docs and guides on OCN about vmoding.
I have how ever never had a problem changing caps so long as they are the same value and volts. You can get some really screwy results if you replace by just volts and not "f"
I'm also not sure that "It is always safe to raise the voltage rating of a capacitor" in all the years the other techs i have worked with, have worked with electronics they have never mentioned that before. In fact i have been told the opposite and to stay with the same values but, that it isn't impossible and in some cases you can upgrade caps just like VRMs.
No need a for flame war. I think people don't discuss the nuts and bolts very often and i find it refreshing when they do.
The voltage rating on a capacitor is just the maximum voltage it can handle without degrading very quickly (= bursting out in flames and smoke).
So it is better to replace a cap with one of the same capacitance value (µF) and a higher voltage rating. If the cap is only used as voltage stabilizer then you can usually increase the capacitance with no ill effects. ESR should always be as low as possible, no exceptions.
wrote:ESR should always be as low as possible, no exceptions.
Some designs do require a maximum series resistance for the capacitor, and in such case, simply swapping out the capacitor for another of the same capacitance and voltage may not be sufficient. I have, however, used a 2-ohm capacitor in designs which call for a 1-ohm and have not had any issues. One of the main issues with old capacitors is that the ESR (equivalent series resistance) often increases with age, to the point where the circuit will no longer function properly. What's worse is that the capacitor may not even bulge.
I'm not aware of any designs which require a minimum ESR to operate correctly. Has anyone seen this, and if so, which type of circuit was it in? I have seen a specification sheet for a mini switching voltage regulator requesting 1-ohm, instead of a "maximum of 1-ohm", but I'm not sure if I can draw any conclusions from that.
If you do not have an ESR meter to test with your bin of old capacitors, I usually try to find a capactir of the same brand, voltage and capacitance, however that is for used caps. If you are buying new capacitors, some places like Digikey, Mouser, and Newark may specify the series resistance, however sometimes it is nowhere to be found. For your old capacitor that you are replacing, try to find the specification sheet for it and see what the ESR is. I usually try to replace the bad cap with a new one with and ESR at or below what was on the spec. sheet.
I wonder what everyone's experience has been with replacing motherboard aluminum electrolytic capacitors with tantalum, cermaic, or film type capacitors?
Plan your life wisely, you'll be dead before you know it.
I'm not aware of any designs which require a minimum ESR to operate correctly. Has anyone seen this, and if so, which type of circuit was it in?
Well, ESR works in conjunction with the coils to be effective at filtering a particular frequency. So if you go too far off the original ESR, it can shift the effective frequency of the filter and end up increasing the ripple on the DC output. That's not a problem you'd normally run into on motherboards and such, since you're generally dealing with very low ESR values to begin with (i.e., going from 0.004ohm ESR to 0.002ohm in a VRM circuit isn't going to change the effective frequency of the filter a whole lot), but it can be an issue in power supplies... on those you generally want to stay as close as possible to the specs of the original caps.
As for replacing caps with different types... sometimes you can use solid polymer caps in place of electroltyics in switching-type VRMs on motherboards and whatnot. I've done it a few times, with quite good results. They're generally lower capacitance and voltage ratings for a given size, though, so you pretty much have to take it on a case-by-case basis whether it's OK to substitute them... it's not something you can just do blindly.
Tantalums could also sometimes substitute for electrolytics, and they're extremely low ESR... in fact, you can occasionally find them in computer hardware, usually '80s stuff, but I know there were a couple Gateway PII/PIII boards in the late '90s that used 'em. There are two problems with tantalum caps though: for one, they're considerably more expensive than electrolytics or polymers, and their technical advantages aren't really large enough to be worth the cost difference. The other issue is their failure mode-- when they go bad, they tend to short out and explode. And I don't mean that figuratively... they sound like a firecracker when they go.
You can't normally replace electrolytics with film or ceramic caps... there's a few problems with doing so, but the primary issue is that you wouldn't be able to get them in sufficient capacitance for the average value you'd find on a motherboard, and even if you could, they would be extremely huge and extremely expensive.
WOW - this turned into quite a discussion 😀
One of the reasons I like the Epox bor versus the Tyan is that the CPU socket is rotated 90 degrees, and there is a lot more room around it, making it easier to find a cooler.
I *thought* about replacing the caps, but right now there's no real reason to.
Also, niter board is in use. I just stole the case from the Tyan system to build an Athlon system. Probably neither board will be used again unless I get nostalgic about K6-2/K6-3...........
Old Thrashbarg, good comments.
Aside from ESR and capacitance values, I have never really gotten deap into the physical understanding of why some designs call for ceramic, electrolytic, film, tantalum, etc types of capacitors. You have raised an interesting point on the modality of failure though, however I have seen electrolytic caps pop and burst into flames on a 386 motherboard before - in fact, that is the only time I've ever seen that happen and was probably a bit of a anomaly. According to specification sheets, usually electrolytic or MLCC caps are suitable for switching regulators. For smaller non-production projects, I have not been deterred by the cost of tantalum caps for values less than about 470 uF. The cost of cermaic capacitors up to about 100 uF is usually acceptable as well, at less than $0.3 each in bulk, or $0.6 for singles, which again, hasn't been an issue for prototype engineering or one-time designs/installations. I've never sourced film caps.
For the issue with DC ripple in regulators, I suppose you could place a series resistor in place with a very low ESR capacitor to see what maximum value ESR your design can tolerate. If your budget can cope with the $7 cost, I have found that using a Recom switching regulator has simplified a lot of these old linear regulator designs. The package size is also smaller than what would be required by adding up all the external components required by a linear regulator.
Your notion of polymer cap replacements for electrolytics in certain applications is good to know. Is there a general rule for the lower capacitance/voltage requirements when compared with electrolytic caps, or is this something you have been able to verify with vendor-supplied SPICE models and/or with a scope? What is the physical understanding for the lower capacitance tolerance? Reduction in time to store charge?
Again, great comments!
Plan your life wisely, you'll be dead before you know it.
Aside from ESR and capacitance values, I have never really gotten deap into the physical understanding of why some designs call for ceramic, electrolytic, film, tantalum, etc types of capacitors.
Yeah, I'm not even going to try that one... that subject is good for a hell of a lot of reading, plus you also have to filter out all the pseudo-science and superstition from all the audiophile idiots and such.
You have raised an interesting point on the modality of failure though, however I have seen electrolytic caps pop and burst into flames on a 386 motherboard before - in fact, that is the only time I've ever seen that happen and was probably a bit of a anomaly.
Well, the main difference is, electrolytics can fail in such a manner, but tantalums usually fail that way. And tantalums are also much more sensitive to voltage spikes and such than electrolytics, so they can fail in situations where an electrolytic wouldn't.
One other trouble with tantalums is that it can be difficult to find larger values of the through-hole type... most of the ones being used nowadays are surface mount.
There's no hard rule for polymer replacements, but just from my own experimentation and testing on a scope, I think I found a fair ballpark guideline. This is assuming you're working on a normal modern-ish motherboard using switching VRMs, and that you keep the ESR the same or lower than it was originally... and I think would apply just as well to tantalums as it does to polymers: for the CPU, memory and chipset VRM outputs, you can usually get away with half the capacitance of regular electrolytics, though with the CPU you should try to keep at least 5-6000uf of bulk capacitance (especially if it's a higher TDP chip). VRM input caps are a little trickier, and since there's not as much to gain by converting 'em to polymer, I generally leave those as electrolytics unless there's some really good reason not to.
But I'll say it again: that's not a rule, just my own observations. Best I can say is just play around with it... and get a scope if you don't have one... you don't need anything fancy, 10-20mhz oscilloscopes are dirt cheap on the surplus market, and VRM switching frequencies are a couple hundred kilocycles or less so a 10mhz scope is more than plenty to see what's going on. What you're aiming for is pure DC... as close to a straight, flat line on the scope as you can get.
The theory behind all this is that, while you need a certain amount of capacitance to compensate for load changes (slow changes, relatively speaking), it's the ripple (fast changes, aka, noise) in the line that's the biggest concern. The input side of a VRM has to deal both with load changes on the VRM, and fluctuations coming from the PSU, so capacitance is a bit more important there. On the output side of a VRM, though, it really doesn't take a hell of a lot of capacitance to keep the voltage stable with load, but larger electrolytics are often used anyway because they're the cheapest way to get the ESR and ripple current capacity to where they need to be.
As for changing the voltage rating of a cap, well, you want at least 20-25% headroom over the actual voltage that'll be going through the cap. So you measure the voltage going through the caps to see what's what, and then apply a bit of common sense. The input side of a modern motherboard VRM usually has 12V on it, so you need 16V caps there. But for the output side of the VRMs, you look at the specs. Like, if it's a P4 board, you know the CPU Vcore is always going to be under 2V, so the 2.5V polymers are fine for the CPU VRM output. If it's a DDR1 board, you're going to have at least 2.5V on the memory, and maybe as much as 3V if it's a board with fancy overclocking features... so you don't want to go with less than 4V polymers on that. And so on. (And just in general, it helps to keep in mind: there are only a specific set of voltages going into a motherboard, so any power running through the board is going to be one of those voltages, or a derivative from one of those voltages.)
I really hope the above was all coherent and reasonably correct... I kinda suck at explaining this sort of thing, plus it's 1am and it's been a long day. 😵
Some interesting replacement strategies on the different cap types. I may play with these ideas the next time I need to design an old school regulator. Those Recom switching regulators I mentioned do not require input or output caps or resistors; it is taken care of as part of a complete package, however an additional input cap is needed if your initial source is of very low impedance, such as a battery.
Some personal observations on your comments are provided below.
I definately do not get pure DC out of the linear vrm on my 486, the noise is on the order of 100 mV and has a repeatable pattern. Its been working fine for years though. I was surprised when I hooked up the scope to it a few months back; I can get cleaner/flatter DC from a noisy AC signal just by using a diode, cap, and leaky R.
For me, I decidedd 100 MHz was a good stopping point, and still cheap enough. At above 150 MHz, the eBay specials cease to exist (at least when I was sourcing scopes a few years back). A 100 MHz scope is good for testing the FSB on boards up to a Super7 and a 100 MHz Coppermine. I have found that the results obtained by the eBay special units are close enough, and often indistinguishable to the multi-thousand dollar branded units from Tektronics and Agilent. In general, we use the branded units at work for official business so the client doesn't come back and ask if the no-name-brand is reliable. However, for my lab at home, the eBay specials sufficie. The main issue with the eBay specials is that the software running the scope has a few bugs, which may never be fixed, however I have learned to work around their bugs. One of the more annoying bugs is when you are reading a waveform and you are using the measurement tools, i.e. Vpp, Vave, freq, etc, they work fine in live playback, but when you hit pause, the numbers jump to ridiculous amounts like from 100 Hz to 1054.6 Hz. Another bug is when you want to take a screen capture, it will remove the measurement tool values (Vpp, Vave, etc) from the screenshot unless it is the first image capture since the device has been powered up. Of course, the cursor measurement values are still available and work correctly regardless of live or paused settings, and I trust the cursor measurement tools more than the automatic Vpp, Vave, freq, etc type of tools anyway.
I have found that SMD caps and resistors can be hand soldered fairly easily, with the proper Weller tips. You can also turn them into through-hole packages if you attach wire leads, but do not put too much torque or shear onto these leads as the solder pad will rip off. You may also be able to epoxy encase the SMD-to-through hole conversion for improved structural strength.
As you pointed out, most of the ripple should be removed by the PSU already. I've designed some simple AC to DC circuits for use in creating various induction sensors, and for simple designs, the desired flatness of the DC voltage can be a trade-off with the initial charge and fall time, as dictated by the RC time constant, and how flat you require the DC output to be. If you cannot find a happy ground for the application, I find plopping one of those $7 Recom units onto that base design flattens out the DC to levels acceptably.
Hope this information is helpful to someone who stumbles onto this posting. I wasn't expecting this topic to become so interesting!
To respond to the original posting as to the differance between Tyan and Epox, you can always improve on either design if you have and SMD rework station and don't mind your motherboard looking like Frankenstein, however this will be a lot more trouble than it is worth. I've always liked Tyan boards, however for the boards I have, Tyan never really seemd to issue BIOS updates. Did they design it right the first time? I dunno. I don't really like how each of their manuals (for different motherboards), walks you through all the same steps of how to install RAM, how to install the CPU, etc. It ends up taking up 80% of the manual.
Plan your life wisely, you'll be dead before you know it.
I see using a higher rated cap (voltage) the same as replacing a resistor with one that has a higher power rating. It can't do any harm...
wrote:I see using a higher rated cap (voltage) the same as replacing a resistor with one that has a higher power rating. It can't do any harm...
As pointed out previously, it depends on if that higher voltage rated capacitor has a significantly different equivalent series resistance as the one being replaced and if the design of the circuit can tolerate it. For many applications, the difference may be tolerated without issue, and in some cases it cannot.
I've seen, for example, 33 uF caps with a 10 ohm series resistance and some with less than 1-ohm series resistance. As Oldthrashbarg pointed out, ensure the design can handle the difference in impedance. Otherwise, if you want to spec a higher voltage capacitor as the replacement, you should be safe if you find a replacement of the same ESR, and in most cases, less ESR, but not usually much more.
Plan your life wisely, you'll be dead before you know it.
There's a lot of valuable information on capaictor replacement in here. It may be appropriate if a moderator could create a new thread with the relevant comments moved over to that new thread.
Plan your life wisely, you'll be dead before you know it.
wrote:WOW - this turned into quite a discussion :) […]
WOW - this turned into quite a discussion 😀
One of the reasons I like the Epox bor versus the Tyan is that the CPU socket is rotated 90 degrees, and there is a lot more room around it, making it easier to find a cooler.
I *thought* about replacing the caps, but right now there's no real reason to.
Also, niter board is in use. I just stole the case from the Tyan system to build an Athlon system. Probably neither board will be used again unless I get nostalgic about K6-2/K6-3...........
I checked my Super 7 boards and the Epox board I have doesn't appear to have more room around the socket then any of the other boards though.
Usually this isn't a problem with s370 and later boards though, they usually left more room around the socket by then.
Well, I was able to get away with one those Nexus axp-3200 coolers on the Epox bord, it would NOT go on the Tyan board 😀
