OpenServo.com

Guest

A common question is whether the OpenServo 1.1 hardware can drive more powerful servo motors -- such as those in the HiTec HS-475 servos. Well, the answer at this point is it doens't seem so. The stall current is outside the specs for the MOSFETs used within the current version of the OpenServo.

An obvious solution is to find different MOSFETs that can handle more current. However different MOSFETs may be difficult to fit within the small confines of a servo.

Another solution would be to double up the MOSFETs in the OpenServo to drive more powerful motors. MOSFETs have the useful characteristic of being able to be placed in parallel to drive more current. With some redesign of the OpenServo PCB a single board could support both lower power motors and higher power motors. For lower power motors four MOSFETs would be used on the board and the other four spots for MOSFETs would be left empty. On higher power boards, all eight MOSFETs spots would be used.

I think such an approach should work in theory. Does anyone have actual experience with using MOSFETs in this manner?

-Mike

nick_A · Joined: 03 Jan 2006 Posts: 14

ive seen mosfets stacked before, this way you have no need to redesign the board. although heat might become the bigger issue

nick

andylippitt · Posted: Tue Jan 03, 2006 4:30 pm Post subject:

How about this guy:
http://www.fairchildsemi.com/ds/FD/FDN338P.pdf

Digkey has a ton of em sold in singles.

mpthompson · Joined: 02 Jan 2006 Posts: 650 Location: San Carlos, CA

Very nice find. These MOSFETs would certainly be worth a try as a direct replacement to power a more powerful motor. When doing testing, keep an eye out on the temperature of the MOSFETs as some heat sink may be needed at higher currents.

-Mike

andylippitt · Posted: Mon Feb 06, 2006 8:59 pm Post subject:

in addition to the MOSFETs going out of spec around 1A, so does the current sense resistor. The current one is a 0.1 omh 0.1 watt which is 1A.

andylippitt · Posted: Mon Feb 06, 2006 9:18 pm Post subject:

Here are some alternative higher current parts I've found. Hopefully someone how knows more than I do about electronics can confirm they are OK.

MOSFETs sold by Digikey in singles:

NDS355ANCT-ND NDS355AN
http://rocky.digikey.com/scripts/ProductInfo.dll?Site=US&V=261&M=NDS355AN
MOSFET N-CH 30V 1.7A SSOT3

FDN338PCT-ND FDN338P
http://rocky.digikey.com/scripts/ProductInfo.dll?Site=US&V=261&M=FDN338P
MOSFET P-CH 20V 1.6A SSOT3

0.5W Current sense resistor sold in 10s:

RP16S.10FCT-ND RP1608S-R10-F
http://rocky.digikey.com/scripts/ProductInfo.dll?Site=US&V=408&M=RP1608S-R10-F
RES HP .10 OHM 1/5W 1% 0603 SMD

dsg123456789 · Joined: 03 Jan 2006 Posts: 21

A 2.2A N-channel MOSFET in the same form factor
http://www.digikey.com/scripts/dksearch/dksus.dll?Detail?Ref=242555&Row=373942&Site=US

A 2.6A P-channel MOSFET in the same form factor
http://www.digikey.com/scripts/dksearch/dksus.dll?Detail?Ref=243586&Row=589662&Site=US

Hitec 422s, which output less torque than 475s, draw up to 2 amps. I think that the board will probably need to be redesigned to accomodate a larger h-bridge, possible one in a through-hole IC. I suspect that 475s and 645s could draw up to 5A max. This week(end) I think I am going to design a new circuit for the openservo that will be able to support these larger servos.

andylippitt · Posted: Tue Feb 07, 2006 1:28 am Post subject:

David, can you provide part numbers... those links are no good.

I think I found your 2.2A N-ch FDN337N
and perhaps the 2.6A P, but the P is not stocked...

dsg123456789 · Joined: 03 Jan 2006 Posts: 21

They are FDN337N for the 2.2A n-channel mosfet and either FDN304P or FDN306P for the 2.7A or 2.6A p-channel mosfet, respectively. I am planning, however, on designing new boards for the Hitec 645 servos, and in those I will probably use a 5A motor driver or something else capable of supplying adequate current.

mpthompson · Joined: 02 Jan 2006 Posts: 650 Location: San Carlos, CA

For those looking to upgrade MOSFETs and needing more board real estate, you may want to look into whether the flyback diodes in the OpenServo circuit are really needed. Being a conservative designer, I added them because I see them in many other h-bridge designs. The flyback diodes built into MOSFETs may be tough enough that the four external ones in the circuit may not be needed.

-Mike

prolinuxfan · Joined: 17 Feb 2006 Posts: 21

Be careful when selecting your replacement MOSFETs, as the Rds-On value is not the only limiting factor in how much current one can put through them. The package form factor is usually the limiting factor, since many times the package cannot dissapate heat quickly enough to keep the junction temps within specs at the higher current levels, and thus, the device will self-distruct long before the theoretical limit imposed by the Rds-On value implies.

Heat will be your biggest problem, and the best way to cut that down is find parts with the absolute lowest Rds On values, so that they produce less heat in the first place.

There is another forum, a Yahoo! group for Open Source Motor Controllers that has much information that may prove useful to this purpose, as the current official Open Source Motor Controller is a single channel H-bridge that uses the principal of parallel MOSFETs to combine 16 of them in a controller capable of 160 amps continuous, and peaks much higher (probably most used in battle-bots).

Could be useful for some of this thread too?
_________________
Allen
(My R2 *WILL* climb stairs when I'm done)

mpthompson · Joined: 02 Jan 2006 Posts: 650 Location: San Carlos, CA

I still have yet to identify the discrete MOSFETs used by HiTec in their high-torque HS-645MG servo as described in this thread:

http://www.openservo.com/forums/viewtopic.php?t=51

I would assume that HiTec chose these specific MOSFETs as a good compromise/balance between the various factors that going into choosing such devices for a servo. The MOSFET package seems to include a soldered tab specifically for dissipating heat into the PCB. It seems logical that using these MOSFETs would seem to be a good starting place to start the next revision to the OpenServo design around.

-Mike

prolinuxfan · Joined: 17 Feb 2006 Posts: 21

One of the other items to keep an eye on in any redesign of the board is how much copper will be available to act as a heatsink for each connection of the FET to the PCB.

It might be a good idea to check that unknown board to see how much copper they chose in square cm to allow each soldered leg or tab, as it would make a difference in the choice of FET too.
_________________
Allen
(My R2 *WILL* climb stairs when I'm done)

ginge · Posted: Sun Feb 19, 2006 1:30 pm Post subject:

This is a very good point.

I actually came across the heating problem fairly early on, and attached a small aluminium (aluminum) plate over the FETS.
This is a hack, and the board will need provisions for heat dissipation

prolinuxfan · Joined: 17 Feb 2006 Posts: 21

Examination of the OpenServo printed circuit board design shows that you may have limited your thermal dissipation by removing all but the thinnest traces, and the existing schematic and parts may even work with the higher power servos with better PCB heatsinking.

I tried to draw up an example of how to alter the v1.1 pcb, but I can't seem to figure out how to attach the gif (5kb) to this message. There seems to be a great deal of room on the board that could be returned to being copper, with a few traces re-routed to add quite a bit of heatsinking to the current schematic.

How do you attach pics?
_________________
Allen
(My R2 *WILL* climb stairs when I'm done)