Active pre-charge/inrush control

The Spark Switch looks exactly like my circuit, only drawn differently. Hmm.. wonder where he got that idea?

For your switch, the LED has separate leads. Wire the LED through a resistor to the load side of the circuit (controller) so it lights up whenever the controller has power. The resistor depends on the pack voltage.
 
he refers to this thread in his github repo for this circuit, so yes :)
 
Does the wattage of the 12V Zener diode matter? Mouser has them from 500mW up to 5W.


Also, which would be a better MOSFET to use for a 12S (50.4V) 60A application? In the early pages of the thread, the MOSFET specified was IRFB4110. However, in Vedder's implementation of Richard Fechter's design, he specifies IRFS7530 in the schematic, which has a higher current rating and lower voltage. If I used IRFS7530, would I need to adjust any resistor values?

http://www.mouser.com/Search/ProductDetail.aspx?R=IRFB4110PBFvirtualkey57370000virtualkey942-IRFB4110PBF

http://www.mouser.com/Search/ProductDetail.aspx?R=IRFS7530TRL7PPvirtualkey51120000virtualkey942-IRFS7530TRL7PP


s0x0td.jpg
 
I used 1W zeners.

Either of those would be fine for 12s. A lower Rds will generate less heat. Nothing else needs to be changed.
 
fechter said:
I used 1W zeners.
Either of those would be fine for 12s. A lower Rds will generate less heat. Nothing else needs to be changed.
the main advantage of the 3077/4410 is the case, as it's easier to install a heatsink to it. and that's the only advantage.
the 7530 are much better for what we need. depending on the number of FETs you use, and the current you plan to run it should be easy to run them w/o heatsink at all.

2x 4110 and 50A will give you 2.3W heat per FET
2x 7530 and 50A will give you 0.9W heat per FET (which should be ok w/o heatsink)
 
izeman said:
the 7530 are much better for what we need. depending on the number of FETs you use, and the current you plan to run it should be easy to run them w/o heatsink at all.

2x 4110 and 50A will give you 2.3W heat per FET
2x 7530 and 50A will give you 0.9W heat per FET (which should be ok w/o heatsink)

Yep.
Here's a quick run at 38degC (100degF) with the Parallel FET Evaluator from an earlier post:

IRF7530_2FET_NoHeatsinkPower.png
Green entries are okay at the rated current for that row (<85degC), yellow = so-so, red = smoke (>125degC).
You can adjust these operating limits in the spreadsheet if you wish (in the column headers in the parts spec table) and the other tables will adjust RYG colors appropriately.

Heat-wise, the advantage over the 3077 is apparent - if you can afford the lower operating voltage...
 
Thank you for all your help guys!

I searched around for other MOSFET options and found some suitable alternatives. 7730 has identical specs to the 4110, but lower Rds and lower price, in exchange for 75V max instead of 100V. I'll probably go with this since I don't need over 60V.

However, the 7537 is attractive for 12S applications because of the price; it's less than half the cost of the 4110 or the 7530! The downside is it has only 230 W power dissipation vs 375 W.

These are the 3 best alternatives I found: (1st = original 4110, 2nd = FET used in Vedder Switch). Note: I only looked at Through Hole mounting because I won't be using a PCB.

Screen_Shot_2016_10_06_at_11_21_37_AM.png
 
teklektik said:
Here's a run showing the current to parallel FET map with your new options:



Wow, thank you! Based on these charts, I see no reason not to use the 7537. Not only is it cheaper, but it outperforms the others (except for the 7530, but I am avoiding surface mounts). My BMS maxes out at 60A, so it looks like 3 parallel 7537's would work, but 4 would be ideal.
 
fechter said:
The power dissipation rating of the FETs doesn't matter much as we want to avoid having them in the dissipation mode.
Hmm - not 100% sure I follow.
In the parts spec table: the max continuous drain current ID(max) is an FYI - really not used, but sort of interesting.
RΘJA is used to find the temp and bracket it in GO/MAYBE/NOGO operating temp ranges and so develop the table coloration.

If you pull down the spread, you can enter any part you want in a row of the part spec table and the other four tables will display the results...
Adjust also: min number of FETs, ambient temp, and "W@°C" temps to define operating temp ranges.
 
In this application, the FET is not trying to dissipate much, so we should not be getting even remotely close to the maximum dissipation ratings, so they don't matter that much. The Rds matters a lot as does the voltage rating. Most TO220 packages are limited to 70A by the legs so the silicon rating just has to be higher than that for continuous but it's nice to have a bunch of headroom for transients.
 
Apologies, I know you are trying to tell me something, but I'm not getting it. I believe the charts already reflect what you are saying - the max ratings are not used anywhere and the dissipation values in the tables are developed from Rds(on) and necessary to evaluate acceptable no-heatsink operation.

fechter said:
In this application, the FET is not trying to dissipate much, so we should not be getting even remotely close to the maximum dissipation ratings, so they don't matter that much.
The Rds matters a lot as does the voltage rating.
Yes - I noted that above - the Id value for silicon in the lower parts table is an FYI - it's not used anywhere. The Vdss is not used either. They are there simply to put FET stats that are commonly called out as significant in handy view so that posted snaps of a chart give the whole FET picture. So, those common specs are unused, but harmless.

The entire chart is based on the Rds(on) spec. The dissipation values in the tables are derived from applying the fractional part of the battery current in the first column to the Rds(on) value according to the number of FETs - a simple approximation barring layout factors. The resulting dissipations are as noted in the table cells and clearly far below the max. These are compared to the dissipation values calculated in the parts spec table at 85/125 degC to determine the GO/MAYBE/NOGO state (GYR color).

I did see that the first column in each of the FET tables was labelled 'Id' which was arguably incorrect, so I changed the column header to the more appropriate 'Ibatt' both in the post above and in the downloadable zip file. Perhaps that made things unclear...
 
jmasta said:
However, the 7537 is attractive for 12S applications because of the price; it's less than half the cost of the 4110 or the 7530! The downside is it has only 230 W power dissipation vs 375 W.

Your charts are great. I was simply trying to respond to the comment above about dissipation rating. I'd say either of those would be OK..
 
Thanks Rich. My Bad. Comments were sort of a smooshed into a continuous stream and I misunderstood...

jmasta said:
My BMS maxes out at 60A, so it looks like 3 parallel 7537's would work, but 4 would be ideal.
Exactly. The charts show still-air lab results so if the unit is exposed to free air flow or has a bit of a heatsink, the FET count could be cut back - but I like 4 myself. As a side note, if you look at the dissipation for the 7537 @ 60A (worst case), the four FETs will be dumping 4 x (0.74W) ~= 3W of heat near 85degC. This could get a little toasty, so give a bit of thought as to where the FETs get mounted. Not much of an issue if you don't run maxed out, but worth a thought...
 
Big electronics dummy here. I have been reading and rereading this thread for several days and I now have a couple of stupid questions.

First Question:
If one looks at the current diagram 'Automatic Precharge 3b' there are three basic conditions: ON, Transit , Off
What I do not understand is that in the OFF condition why is the second 1M Ohm resistor needed. When the switch is closed you have a direct path from the gates to ground. Does that not pull everything in the middle to ground thus shutting off the gates ??? I could understand if the two 1M ohm resistors were needed as a voltage divider but I thought that the first resistor and the zener diode would fulfill that role.
Question 1 - edited(400).pngView attachment Question 1 edited.pdf
Second Question:
Could not the switch be moved to the junction between the two 1M Ohm resistors? Would this not reduce the current flow in the off condition by interrupting the path from first resistor to ground? ... and in this condition what purpose does the second 1M resistor serve as you have a direct path to ground ???
Question 2(400).png

Edited:
View attachment wire diagram question edited.zip
Question 1:
added note to lower 1M ohm resister in "ON" condition
Revise switch ... it was drawn incorrectly
Added optional fuse

Question 2:
Remove lower 1M ohn resister
Revise switch ... it was drawn incorrectly
Added optional fuse
 
Edited for correction:

You are correct. The lower 1 meg resistor is not really needed in that configuration. It only serves to make sure the gates are off if the switch is left on and the battery gets disconnected. This can happen if your BMS trips, for example.
 
Hi all,
I would like to build or buy (if someone can) the circuit, my application will be a 12s 80A constant and 150 peak (5 seconds) and my intend is only prevent the spark on the connector.
I don't need the switch so I'll use the scheme (https://endless-sphere.com/forums/viewtopic.php?f=3&t=40142&start=125#p712738)
The best is mount it closer to the battery or to the controller?

If I am not wrong I need this parts:
IRFS7530 or IRFP4468 FETs (4 or 5 pieces?)
1M ohm resistor
1K ohm resistor
1uF Capacitor
2W 12V Zener Diode

Thanks for all the thread, is very helpful even if I'm a newbie.
 
did some testing of the active pre-charge. for my project i need one to protect the main contactor from damage. i can't use a simple resistor because the dc/dc converter is using some current.

first i build up the "Automatic Precharge 3" with a switch an it worked great.
Automatic Precharge 3.png
nice liniare voltage rise and constant current.

next i removed the switch and got this result.
Automatic Precharge 3_no_SW.png
i think the high peak current is caused by the capacitor voltage divider made up by the cap in the controller and the 1uf(c1) in the circuit and the gate of the mosfet. the caps are charged through the resistor R2. the gate has the smallest capacitance and will be charged the quickest.

by adding a 10uf cap across the the zenerdiode, the voltage divider changed and gave this result.
Automatic Precharge 3_10uf.png
the constant current is back, but it takes several seconds before it turns on.

to fix it i added a mosfet to the circuit.
Automatic Precharge changed.PNG
when the battery is connected the Q2 is switched on before the main mosfet Q1. and shorting out the gate of the main mosfet just long enough for the capacitor c1 to charge up. (c2 is not used) and it works nicely.

Automatic Precharge autoreset.png
added bonus, when the battery is disconnected and then reconnected before the circuit is discharged Q2 will reset the system.

for this test i used a 8800uf 50v capacitor
 
Good stuff. I see they just use a diode, resistor and capacitor to take care of the startup spike.
 
Back
Top