Instant Start 18 fet Infineon Boards are here...

[methods]

Right, but 24S A123 is only 80V nominal.
I think the original discussion was revolving around how Doc is pushing for 160V caps
I was arguing that 100V caps are plenty for most people, you included.
I am even running 24S Lipo on the 100V caps without issue (yet) and these are the caps from China.

You will get better performance with more capacitance at 100V than by bumping up to 160V and losing capacitance.

I agree with you though, no point making anything less than a monster.
I really dont think we need to worry though. I dont think anyone here (on a bike) can draw more current than this controller can handle.
If we get into the motorcycles... Then maybe we will hit some limitations - but my guess is that they will be practical limitations, - like wire gauge.

-methods

[Doctorbass]

Right, but 24S A123 is only 80V nominal.
I think the original discussion was revolving around how Doc is pushing for 160V caps
I was arguing that 100V caps are plenty for most people, you included.
I am even running 24S Lipo on the 100V caps without issue (yet) and these are the caps from China.

You will get better performance with more capacitance at 100V than by bumping up to 160V and losing capacitance.

I agree with you though, no point making anything less than a monster.
I really dont think we need to worry though. I dont think anyone here (on a bike) can draw more current than this controller can handle.
If we get into the motorcycles... Then maybe we will hit some limitations - but my guess is that they will be practical limitations, - like wire gauge.

-methods

Het Methods, I just bought 50$ of high high quality Panasonic HC capacitors (-40°C ~ 105°C).. still 1000uF and 160V and low ESR with 22mm x 40mm size.. it's 4mm larger diameter but they will fit!

so i'll have more capacity.. i'll replace the actual 470uF for 1000uF and will still have 1000uF for the 1000uF.. so it will be 1060uF more for a total of 5000uF at 160V instead of 3940uF at 100V... operating capacityr without safe voltage margin will make them to age faster and their ESR will increase as they age.. that is something to know .. it's similar to battery aging.



Doc

[methods]

How much more expensive are the 160V caps than the 100V caps?
If they are similar price and they fit as well as you say they do then I will consider them
I can tell you right now though.... 4mm is a lot!
Currently it is a tight fit with all that 10 AWG wire in there.

I suspect that you will get interference with the 10 AWG and it will cause you to have to route around them instead of over them
(which I guess is ok)

-methods

[Doctorbass]

Here is what solution i found :

[steveo]

Here is what solution i found :

I used 3 1000uf 200v capasitors in my 133v controller!!

my diy controller still running strong

-steveo

[methods]

Here is what solution i found :

eh eh... Not so fast Doc
You forgot a few things.... Like the 3 giant phase wires
Also you will find that the bottom right of the board has heavy congestion due to all the control wires being grouped right there.
Also the VCC comes from up on the left corner.

I am not saying that it can not be done... I bet I could put 6 AWG wires in there and 600V caps.
Doing it once is one thing.
Building 30 of them is another.
Convincing a fabrication shop to take the job..... That is yet another.

Usually i think like you Doc.
On this project I am being forced to think about compromise, ease of build, repeatability, etc.
I have built many high consequence systems at work and I have learned that tight cable routing leads to great frustration and problems.

But I dont want to argue... If the build works out good then please share with me the part numbers and I will try them out.

-methods

[dnmun]

i used the 100V1000uF caps on my 72V conversion of the 9 FET infineon: viewtopic.php?f=2&t=10222

the caps are 18mmx40mm long so i had to move the power resistors to the underside so the two on the input would have space inside the case.

i am selling the caps i got on ebay for 30 cents apiece plus postage. i need to put up a sign in the for sale section yet. i also have 5 of the 100V2200uF caps i might part with, but you have to have a need since i am gonna use them on the S/D busses when i do the next upgrades to 72V.
nichicons.

[Doctorbass]

Oh...

digikey is SO FAAAAAAST !!!

I received them at the diner !!! less than 20hours after my order!!!!

12 brand new 160V 1000uF caps...

Methods I agree with you that building 30 controlelrs with biggers caps is not as simple as it appear to be !!!

but you know me.. for building one or two.. i'll find the place!!!

I will need to use a precharge resistor otherwise i will blow the switch after 10 or 15 power ON action!!!

The 18 mosfet Crystalyte is on the way too.. i will have both and will compare them after i'll upgrade them.

ok.. now i quit.. i must go to assemble the aluminum case of my mini A123 killacycle pack !!!.. it will be very similar to 1/10 scale of it


Doc

[liveforphysics]

I just had a realization about this cap stuff.

We seem to be going about it the wrong way. We aren't concerned at all with storing energy in these caps, that's why we have batteries. We have the caps to resist the change in voltage each time a FET bank switches on or off. The peak capacitance we need to be store to do this is under 50uF (from my crude estimation).

All the effects our performance is the final ESR of the system of caps we use. In other words, any cap solution that provides a lower combined ESR, regardless of the capacity values is going to result in higher performance.


This reminds me of my car audio days, when I was able to keep my amp voltage more stable with a single 1/2F cap than with the 16F cap bank that I had bought thinking it would be an upgrade. It used those carbon supercaps inside, which have crap Ri, and it didn't help keep voltage stable any better than adding an additional little wimpy 12v SLA battery would have done.

[Doctorbass]

ok guys.. let's show some real data

This afternoon i've tested the capacitors ESR and capacity using a high quality laboratory grade HP LCR tester ( 15000$)

here is the data.. ( i thing you might consider replacing those stock caps)

Infineon 1000uF 100V Rubycon: cap: 998uF ESR: 110mohms
Crystalyte 1000uF 100V DTDZ : cap: 1063uF ESR: 260mohms
Panasonic 1000uF 160V Panasonic:cap: 740uF ESR: 73mohms

It seems that buying these 4.50$ caps is the best thing to do for high power/voltage controller!

Doc

[swbluto]

I just had a realization about this cap stuff.

We seem to be going about it the wrong way. We aren't concerned at all with storing energy in these caps, that's why we have batteries. We have the caps to resist the change in voltage each time a FET bank switches on or off. The peak capacitance we need to be store to do this is under 50uF (from my crude estimation).

All the effects our performance is the final ESR of the system of caps we use. In other words, any cap solution that provides a lower combined ESR, regardless of the capacity values is going to result in higher performance.


This reminds me of my car audio days, when I was able to keep my amp voltage more stable with a single 1/2F cap than with the 16F cap bank that I had bought thinking it would be an upgrade. It used those carbon supercaps inside, which have crap Ri, and it didn't help keep voltage stable any better than adding an additional little wimpy 12v SLA battery would have done.

Which is why replacing more capacitors in parallel should only help, as more effective series resistors in parallel means a less total effective resistance. However, lower ESR caps should help more if you're only changing out caps.

By the way, I didn't know ESR was such a big deal for caps. What kind of failures do high ESR lead to? Higher probability of cap failure due to over-heating? A slowing down in "capacitor response" time due to the higher RC time constant which leads to...? Any others?

[fechter]

here is the data.. ( i thing you might consider replacing those stock caps)

Infineon 1000uF 100V Rubycon: cap: 998uF ESR: 110mohms
Crystalyte 1000uF 100V DTDZ : cap: 1063uF ESR: 260mohms
Panasonic 1000uF 160V Panasonic:cap: 740uF ESR: 73mohms

What frequency were you doing the ESR test at?
I would suggest testing at 10-15khz and also at 1khz. 15khz will be near the PWM switching frequency and 1K will be closer to the commutation frequency.

You may be surprised with the results of of lower capacitance samples too. Try a 470uf or even a 220uf and see how it compares

[Doctorbass]

here is the data.. ( i thing you might consider replacing those stock caps)

Infineon 1000uF 100V Rubycon: cap: 998uF ESR: 110mohms
Crystalyte 1000uF 100V DTDZ : cap: 1063uF ESR: 260mohms
Panasonic 1000uF 160V Panasonic:cap: 740uF ESR: 73mohms

What frequency were you doing the ESR test at?
I would suggest testing at 10-15khz and also at 1khz. 15khz will be near the PWM switching frequency and 1K will be closer to the commutation frequency.

You may be surprised with the results of of lower capacitance samples too. Try a 470uf or even a 220uf and see how it compares

i've used 10kHz.

yes that could be interesting to test lower value!

Doc

[liveforphysics]

It's great to know that a single one of those panasonic caps would help more than 4x the crystalite caps! That's quite a suprize, because generally ESR increases directly with voltage rating for capacitors of the same technology (electrolytic in this case).


I know there are 12v rated caps used for computer MoBo's that are 2mOhm ESR. It would be fantastic to find an ultra low ESR cap for voltage ratings closer to what we are working with.

[CamLight]

By the way, I didn't know ESR was such a big deal for caps. What kind of failures do high ESR lead to? Higher probability of cap failure due to over-heating? A slowing down in "capacitor response" time due to the higher RC time constant which leads to...? Any others?

Those are the big ones.
High temperature is death for caps. All that charging/discharging through that equivalent resistance creates a LOT of heat. Lower ESR means both faster response (if needed) and cooler running caps. Electrolytics have a short enough life span already without heat making it worse.

However, there are some voltage regulators that require a minimum ESR to remain stable for certain voltage/current operating ranges.

[liveforphysics]

I found this site, it's kinda neat.

http://www.low-esr.com/

Too bad I didn't find anything too fantastic for our E-bike applications. If you were running a 24v system, going for the 35v hybrid caps would be great, they are extremely small so you could package one on the board locally for each FET.

But, nobody runs 24v that is interested in power, and grouping them in series would end up with a lot of hassle for a minimal performance gain.

[liveforphysics]

Rubycon makes some caps that look pretty good, but you would need 3 in series... which would be lame to setup.

35v 470uF 0.014Ohms ESR 10mm x 20mm

So, it would be 42mOhms for a 105v cap with 156uF, and it would fit in a very small space. I think using a group (3s8p) of these caps you could get a board down to maybe 5mOhms.

Running 4x of Docbass's panasonic caps would get you to 18mOhms, and it would be a lot easier to solder up


I'm going to keep hunting for the ideal cap. I found ultra low ESR caps in voltages up to 200v, but they maxed at 1uF. Really impressive specs though! I just need to find a larger size of cap using the same polymer technology rather than the regular cheapy liquid electrolyte style.

[methods]

Keywin forgot to populate D32. This is the diode that links the -EBS (ebrake) pad trio. It is suggested that you use a 1N4007 or an M7.

Doc & Gary -> you are on your own
I know it is rocket science but I think you two can manage

-methods

[boostjuice]

Are these physically too large to fit inside the infineon casing?

1200uF, 100v, 62mohms, 1.375" in Diameter x 3.125" in Length, $16.74

http://www.mouser.com/Search/ProductDetail.aspx?qs=sGAEpiMZZMsCnlYck6hSqO6KPCXhPvFm4H11sKdCUik%3d

720 uF, 150v, 63 mOhms, 1.375" in Diameter x 3.125" in Length , $17.16

http://www.mouser.com/Search/ProductDetail.aspx?qs=sGAEpiMZZMsCnlYck6hSqO6KPCXhPvFmXU7gjc7nu2c%3d

[methods]

The 18 fet Crystalyte boards are IDENTICAL to these boards.
They are buying from the same manufacture and just populating with slightly different parts.

http://endless-sphere.com/forums/viewtopic.php?f=2&t=10627&p=163176#p163176

Lol... Poor old Doc is going to have 2 of a kind.

-methods

[Doctorbass]

I found something interesting on the XIECHANG website!! some new indo availlable about the 18 fets controller!! and the uC they use ( crystalyte and Infoineon!!)

here is the 18 fet Infineon page: http://www.xie-chang.cn/cn/product_read_kzq.asp?id=117

you can translate using Babelfish..

This is great to have update on their page about the 18 fets controlelr and the new Uc pdf!

Doc

[wanders]

I found something interesting on the XIECHANG website!!

Yep, all of Keywin's controllers (whether with an Infineon microcontroller or not) come from Xie-Chang (see my post on page 2 of this thread). Their website has a lot of dowloadable documentation, all of which is copy-paste protected (to prevent automated translation among other things, I guess). I have fixed this for all their docs, and another ES member (biketrials) has said that he would help translate. The automatic translators (e.g. Google) don't do a very good job on this stuff. Hope to have some results soon.

On a slightly different tack, methods, are you programming these things using the USB cable? The software is available on the Xie-Chang site (I can point you to it) but it is probably all in Chinese. My controller is still XC846 (Infineon) based, unlike the 18-FET version. I need some pointers on how to use the direct USB cable which came with my Xie-Chang controller (a 12-FET 72V/45A version). For example, do you program with the power on, or does all the programming power come from the USB cable?

[methods]

DUDE - DOC - you are a genius!

I think that is the new software that I have been waiting for keywin to Translate!!!
I can get the translated tomorrow and we will have the software running!

I secured a translator at Stanford University. He is perfectly fluent in both Mandarin and English.

Awesome!
I am going to have to send you a free gift if this is the software that I think it is.

-methods

[Doctorbass]

DUDE - DOC - you are a genius!

I think that is the new software that I have been waiting for keywin to Translate!!!
I can get the translated tomorrow and we will have the software running!

I secured a translator at Stanford University. He is perfectly fluent in both Mandarin and English.

Awesome!
I am going to have to send you a free gift if this is the software that I think it is.

-methods

[Doctorbass]

Something INTERESTING FOR YOU GUYS !!


Here is the complete folder of the XIE CHANG web page concerning controller!!

I translated the folder but not the file content.. ( few hours are needed for that!)

I ziped it for you!!

Alot of great INFO about infineon.. and i guess crystalyte controlelr desing!! and their Uc !!... and some .exe!!

I let you discover how to use these files!

My part of job is done for that.. I must return to my drag bike fabrication!

Doc