Low ESR Caps for a CNC Servo Amp

[arsenix]

I am working on changing out old capacitors on my 1998 era CNC (a Bridgeport Torque Cut 22 VMC). I'm finding that there don't seem to be similar spec caps available anymore, and wondering if folks here have an explanation.

The servo amps in the machine are pretty simple units. The machine has brushed DC servo motors, and there are amp cards for each axis that have a handful of IGBTs that switch ~130v DC at 25kHz. The 130v DC bus is rectified from mains with a large transformer and damped by a massive 15mF 250v cap. The cards have each have a handful of large electrolytic caps on them. I've been noticing some vibration under load and based on looking at the DC bus with a scope I'm fairly certain either the main filter cap and/or these smaller ones are on their way out. I figured I'd swap them all since they are almost 25 years old at this point!

The caps on these cards are 250v, 180uF caps marked 673D. They are marked 105C, have a date code and "UCC" on them (I presume this is United Chemi Con). They are 22.5mm in diameter, 70mm tall, and have 7.5mm lead spacing. 673D seems like some very old capacitor spec shared among multiple companies (or maybe there was some M&A that went on?). I found a few datasheets for these but not the exact ones (https://www.mouser.com/datasheet/2/427/673d674d-1211217.pdf). The interesting thing about these caps is that they have quite high ripple current rating and low ESR. Around 3A ripple from 20kHz to 100kHz, and ESR around 100mOhm at 20kHz.

I figured no big deal do some trawling on digikey, mouser etc and find some replacements. The odd thing is that the only caps I can find are much smaller physically (like 1/4 the volume). They have excellent lifetime ratings (10-12k hours even at high temp), but have significantly higher ESR and lower ripple current than these caps. It seems like to get a similar equivalent ESR I would need 3-4x as many (ie same volume!). I'm wondering why this is. Given the application I think the lower ESR is likely a good thing, although in practice it may work perfectly fine with the higher ESR. Do modern equivalents of these capacitors just not exist?

[strawberry]

larger capacitors should be 90% fine, no need for replacement if parameters are still fine even after 40Y
larger capacitors = longer life , lower ESR
modern capacitors use water based electrolyte = low ESR high temp
if water electrolyte chemistry is not right capacitor generate gases(H/O)(leaking due to pressure) when overloaded/stressed/aged

[tooki]

Well, here’s almost certainly the correct datasheet. My guess is it’s the 673D187F250HP1C. Worst case, write to Chemi-Con and ask for a recommended replacement.

I searched on DK and found some potential substitutes (by similar high-frequency ripple current):
Nichicon LGR2E181MELA25
Kemet ALC70 series e.g. ALA7DA181DC600
Kemet ALA7D series e.g. ALA7DA181CE600

(I searched for “chemi-con 673d” and found this link https://www.westfloridacomponents.com/mm5/graphics/M04/673D107F250.pdf )

[arsenix]

Thank you for that! That does look like the correct datasheet and it lists this specific device.

One thing I realized after digging a bit more is that these devices spec their ripple current at 85C despite being 105C "rated". Newer devices that are rated for 105C spec ripple current at that temperature. They have a table of temperature to ripple current derating and at 85C it is 1.6x the 105C rating. So some of the devices I am looking at do in fact have similar or higher ripple current ratings. This makes sense to me given that they are smaller and likely have better heat dissipation. They still seem to have significantly higher ESR though.

It also seems that the "snap-in" capacitors have lower ESR and higher ripple, which is maybe a modern trend?

The 673D in 180uF/250V has an ESR of 117uOhm @ 20-40kHz,25C. I haven't managed to find almost anything that low. I found a Panasonic cap (EEU-EE2E221) that has a nice high ripple rating (~6A @ 20Khz,85C). With a dissipation factor of .15@250v, that gives an ESR of ~900mOhm@120Hz. The 673D has an ESR of 383mOhm@120Hz. I found a few snap-in caps that low but nothing I could actually buy. I've also been limiting my searches to what is in stock, which these days can be pretty limiting!

[T3sl4co1l]

Capacitors of that value will be best in snap-in styles, with short leads.  Which, I think 7.5mm lead spacing is standard in the smaller ones, there?

If you find [long-]leaded parts of that value, leaving them long and forming to fit is fine.  Add some glue to keep it from flopping around.

Tim

[arsenix]

10mm lead spacing seems much more common for the snap in styles. Think it is conceivable I could bend the leads on a 10mm spacing cap to fit a 7.5mm spacing PCB?

[David Hess]

I figured no big deal do some trawling on digikey, mouser etc and find some replacements. The odd thing is that the only caps I can find are much smaller physically (like 1/4 the volume). They have excellent lifetime ratings (10-12k hours even at high temp), but have significantly higher ESR and lower ripple current than these caps. It seems like to get a similar equivalent ESR I would need 3-4x as many (ie same volume!). I'm wondering why this is. Given the application I think the lower ESR is likely a good thing, although in practice it may work perfectly fine with the higher ESR. Do modern equivalents of these capacitors just not exist?

Modern capacitors require a smaller volume to achieve the same voltage and capacitance product.  With a smaller volume comes less surface area and higher thermal resistance so all other things being the same, they have a lower ripple current rating.

Some or all of the different can be made up by using a larger capacitor with more capacitance or higher voltage or both.  There is a jump increase in ESR at about 160 volts but that does not matter here.  Usually I select a higher voltage part but there appears to be no benefit to doing that here.

I suspect the 250 volt 180 microfarad UCY2E181MHD Nichicon parts or similar parts from Rubycon and United Chemi-Con will work fine despite their lower 1.1 amp ripple current rating.  I doubt the original parts were selected based on ripple current rating or ESR.

[Audiorepair]

10mm lead spacing seems much more common for the snap in styles. Think it is conceivable I could bend the leads on a 10mm spacing cap to fit a 7.5mm spacing PCB?

No.

Drilling an extra hole is a better idea if that can work.

[arsenix]

Modern capacitors require a smaller volume to achieve the same voltage and capacitance product.  With a smaller volume comes less surface area and higher thermal resistance so all other things being the same, they have a lower ripple current rating.

This makes a lot of sense. Interesting how in this case tech progress has driven the specs in a different direction. I suspect you are right that these lower spec'd caps will work. At these voltages, the losses from ESR in the caps are going to be pretty minimal. I suspect these caps were chosen for their temperature rating and long life, given their proximity to the IGBTs and a quite massive heat sink.

Drilling an extra hole is a better idea if that can work.

In this case I think I can do that pretty easily. These are two layer boards and they are pretty wide open. I'll need to install some type of via since the power busses are on both layers, but it is doable.


I ended up ordering some Panasonic caps with 7.5mm spacing that seem to have appropriate ripple current rating, and then also some snap-ins to try that will require modding the board to mount.

[Audiorepair]

Drilling an extra hole is a better idea if that can work.

In this case I think I can do that pretty easily. These are two layer boards and they are pretty wide open. I'll need to install some type of via since the power busses are on both layers, but it is doable.

The unused old hole will still be in existence and providing a via.

You may want to create a pad on the top side of the drilled hole and tin it, so that the solder can wick up the cap lead and solder both planes making even more vias.

[T3sl4co1l]

Hrm... non-plated thru-hole, and you expect it to make connection on the top side?  I don't know about that...

If you can get something to fill the hole, though, that would do.  (There are for-purpose rivet kits out there, but you probably wouldn't bother for just one/few-off.)  Just a bit of, something thin or flat, a lead clinched around, a bit of solder wick, a bit of cleaned PCB foil -- would do well enough.  There's a fair amount of space around those square/flat pins.  Solder it in place, it'll probably stay in contact on top while you're soldering the pin on the bottom. Probably.

Tim

[arsenix]

The unused old hole will still be in existence and providing a via.

Right!

[Audiorepair]

The unused old hole will still be in existence and providing a via.

Right!

Fill it with solder/wire/both.