I’ve done some tinkering attempting to duplicate the H-bridge in the schematic and have been unsuccessful so far in getting it to work as described by the origional builders. Trying to drive it as described from the servo motor leads generally results in a dead short thru the P and N MOSFETs, letting the magic smoke out of my fingers when I touch them. Definately more tinkering to do.
A while back in Nuts&Volts there were articles on using MOSFETs, including using NPN MOSFET transistors as high side drivers. The below was given as a simple way to impliment this approach to bootstrap gating the transistor when the transistor when is being turned on/off faster than ~1HZ. Just wondering if this would be practical in an h-bridge driven by the power pulses being sent to the servo motor. Apparently this is found in switching power supplys.
Went to the Harbor Freight parking lot sale today and picked up one of the winches (below, shown with a standard servo) to tinker with for $39.99 plus tax. I was hoping the control switches would be easy to push, but they are fairly stiff. The idea being that a single servo could operate the switch buttons (eliminate the need for a large H-bridge). Spring pressure in the buttons would need to be reduced. Current operating thoughts would be to use a servo to control the control switches, a pot mounted on the winch to show the drum position and provide an analog value to a controller/pc (via an ssc-32 for the pc), and the controller/pc to send the servo control info to position the button control servo as needed to position the winch. A strong mounting point would need to attached to the winch spool for leg mounting and such. I could see three of these winches being used on a large scale Kronos robotics three servo hexapod, possibly large enough to ride on. Not as dramatic as the spider car, but possibly cheaper and easier. 
Well, why not use relays or solid state switches (even a servo driver), instead of trying to push the buttons with R/C servos!
Also, the winch takes a lot of turns, so a typical pot is out. Just use an encoder, and somehow put limits on the cable payout. This could be done with “knots” (little clamps) on the cable, arranged such that they move a “fork” when the two extremes of cable travel (all out, all in) are reached.
Alan KM6VV
The quick answer is cost. I’ve priced relays and solid state switches and they can’t compare to the possible $5 el cheapo servo price. Idealy the origional heavy duty switch contacts would be used. They have a positive on/off click, which helps prevent a weak switch contact pressure.
I think you missed the point that this would be used like a large servo which rotates ~180 deg max. The cable would be removed from the spool and replaced with attachment hard points.
Yeah, I missed the 180 degree. Then you should be fine with a pot.
You might check for surplus relays. I’d really caution against trying to use R/C servos to push buttons!
How about using the R/C servo (if you must) to tip a pair of mercury switches back and forth? Or move a magnet to operate reed switches? Although they probably won’t handle the current. Automotive relays might be a possibility, but you’d need a transistor driver.
Just ideas.
Alan KM6VV
Just got a hand full of the L298N H-bridge chips (below) and wondering if anybody has any experience using them in projects. I’d like to drive these chips using the board out of a standard servo. One thing I’m not familiar with in his chip is the “fast motor stop” function of the chip. Not sure if this will be an issue in my intended setup. I’ve noticed that the stall current of the standard frigelli linear actuators is ~4A, so this chip might make a decent driver for these when paralled. I’ve also noticed that the really big linear actuators (~1,000 lb.) usually have inline 10A fuses, so several L298Ns in parallel might be able to handle these.
pololu.com/catalog/product/924
pololu.com/file/0J67/L298.pdf
I don’t think one can parallel these chips. They have a bipolar output stage and if memory serves you can’t parallel bipolar transistors for more current. FET’s can do this though…
The datasheet doesn’t say much about the internal workings of the chips, but there are two seperate H-bridges in each chip. Figure 7 shows the two H-bridges being put in parallel operation for higher current needs and is mentioned in the applications section. There is also a scensing pin for each H-bridge that can reportedly used to limit current flow thru the H-bridge (but no details on just how to set this up). My thinking is that if the two seperate H-bridges in each chip can be put in parallel, then more than two may be possible.
It’s been a long time, but I think there is a big difference between paralleling two drivers on the same chip, and paralleling driver chips. I think there has to be serious monitoring of the current to ensure one chip doesn’t do all the work. Good luck with it!
The H-bridges inside of the chip are closely matched, and are allowed to be paralleled. Outside of the chip, you’ll run into problems without complicated current balancing schemes. Better to go to a bigger chip in that case.
Alan KM6VV
I just don’t know of any bigger H-bridge chips. One has to jump from the inexpensive chips to somewhat expensive H-bridge boards. I see servocity has taken a $129 linear actuator and added a $170 gizmo setup to make their $299 servo linear actuator.
Check out the LMD1845.
Google “LMD18245 + “DC motor” + robotics”
3A, 55V
Alan KM6VV
I looked at the LMD18245 data sheet and it has some interesting features, but it doesn’t appear to be easily adaptable as a big servo control device. It also has a price tag of $16 each. I did notice that the drain side (hoping I’m getting the terms correct) transistors appear to be charge pump operated NPN MOSFETs.
er, N-channel MOSFETs.
The technique of using a bootstrap diode and capacitor or an outright charge pump is pretty common with higher power H-bridges. N-channel devices are more generally efficient and cost effective than P-channel MOSFETs. This is even more important at higher currents and if you are using 4-quadrant control where both high and low side devices are chopped (vs. 2-quadrant where just the bottom side devices are chopped.)
I previously posted the below schematic of a high side boot strap setup that apparently is found in switched power supplies and such. My question is will it continously pass current when the driver supplies pulses greater than some minimum frequency. The N&V article mentioned conduction at greater than 1 HZ, but didn’t mention if the conduction was continous or pulsed at the same rate as the driver.