Servo Booster...


 
Bryan Mayland said:
At Radio Shack they usually carry the equivalent 2N2222 part. It is identical except the pins are mirrored.
Click to expand...

I went back and edited the circuit diagram adding the Collector, Base and Emitter designation on the transistor to make it easier to rig up with the 2N2222 or other transistor...
 
I looked for a premade servo booster at any online store and they all seem to be UK stores, oh well. I have the parts to make one, I believe, so will give it a try this weekend. I could even make a little pcb, as I have some Copper boards laying around.
 
John Bostwick said:
I looked for a premade servo booster at any online store and they all seem to be UK stores, oh well. I have the parts to make one, I believe, so will give it a try this weekend. I could even make a little pcb, as I have some Copper boards laying around.
Click to expand...
I found the same, most of the boosters were coming from over seas, and the ones I could find here were kinda pricey ($16 - $20). Kinda ridiculous to order from halfway around the world or pay through the nose for something you can accomplish with a 20 cent transistor. As for the circuit board, I don't really see the need for that either, some shrink tubing over point to point wiring is just fine, it's a tiny little thing... I made mine with exaggerated space between the parts for easy demonstration here.
 
I just tested the booster out at the damper end of the long CAT5 cable... The good news is it stopped the spasms, the bad news is the servo wouldn't move at all, so it's definitely better to put the circuit on the controller end. This stands to reason, cause the damage to the signal is already done at the end of the cable, you want to boost the signal when it's clean not after its degraded and full of interference.
 
I did some testing last night with the oscilloscope and I can't see any signal degradation at all. The longest ethernet cable I have around the house is only 25ft but checking the signal on the far end it is perfect with regards to voltage, timing, and edge strength. I tried both a v4.1 and v4.2 board and they're both the same. I can't test it with the servo load on it on the far end because I can't access the pins once it is all plugged together so maybe that's the difference.

I'm still going to try the 5V booster (4.3V booster is technically more accurate) when I get a chance though.
 
Just some follow-up in this topic...
I've been running my HMv4.0 with the servo booster (picture above in thread) in place for a while now and I have had very solid servo performance, haven't had one episode of the servo spazz'n out at the end of my very long CAT5 cable since I put it in place....
Bryan noted in the Development Log thread that he integrated the servo booster into the next version of the HM board (using a MOSFET in place of the bc337 transistor), so forward looking this booster will be obsolete. If you have a HMv4.1 or older and have issues with servo chatter or need to run the pit over an very long CAT5 cable you might try building this simple circuit into your Heater Meter or into your CAT5 cable....
 
I received the v4.2.3 test boards yesterday and my MOSFET driver wasn't working and as soon as I looked at it I knew it shouldn't work. I popped it out and put in your NPN version and... I'm not sure this circuit does what you think it does.

The NPN forms a "common collector" amplifier circuit if I'm reading your drawing correctly. It's actually reducing the voltage, not boosting it. The drain voltage doesn't matter (as long as it is above the gate voltage), and the output voltage is always 0.5-0.7V below the gate voltage. The theory holds up as my oscilloscope shows 2.8V PWM max voltage with your circuit. I was under the impression that you were trying to boost the servo PWM voltage to 5V.

This circuit would increase the drive current, but I wouldn't expect that to make a difference because the servo input should be high impedance. In short, I'm not sure why your booster works, unless you've got some high resistance connections somewhere along the signal path.
 
You would probably need something like this;

Mosfet_level_shift.png
 
Well, never went deep into theory or measurements after I built it, but it does indeed strengthen the servo and also makes the signal work way out over the CAT5 cable where it flakes out when it is not in place. I have done lots of in/out testing (since I build the handy inline version) and the benefit is evident and reproducable. Not sure about the MOSFET circuit though. Apparently a boost is drive current is what the servo needs...
 
I thought that Ralph changed the power in his circuit to be 5V and not 3.3v. At least that was my impression when I looked at his circuit. So it should be a gate voltage below 5V which would be better than 3.3v from the pi.

dave

Bryan Mayland said:
I received the v4.2.3 test boards yesterday and my MOSFET driver wasn't working and as soon as I looked at it I knew it shouldn't work. I popped it out and put in your NPN version and... I'm not sure this circuit does what you think it does.

The NPN forms a "common collector" amplifier circuit if I'm reading your drawing correctly. It's actually reducing the voltage, not boosting it. The drain voltage doesn't matter (as long as it is above the gate voltage), and the output voltage is always 0.5-0.7V below the gate voltage. The theory holds up as my oscilloscope shows 2.8V PWM max voltage with your circuit. I was under the impression that you were trying to boost the servo PWM voltage to 5V.

This circuit would increase the drive current, but I wouldn't expect that to make a difference because the servo input should be high impedance. In short, I'm not sure why your booster works, unless you've got some high resistance connections somewhere along the signal path.
Click to expand...
 
Yes, I put 5V to my circuit, not 3.3v. The inline version I made is using the +5V that is powering the servo as its power source.
 
Last edited:
It doesn't matter if you put 3.3V into it or 5V into it, you get out 2.8V. It's a function of the gate voltage, not the collector voltage.
 
Peter F said:
You would probably need something like this;
Click to expand...
Yeah exactly except I built it with one resistor because it is unidirectional. Of course, I hooked it up wrong on the schematic so /facepalm.

I'm going to do some testing tonight to see how much current there is running to the servo on the signal line, just for my own curiosity. Ralph, do you have any 2N7000 or BS170 lying around, because I'm curious if the 5V conversion would work as well as the current conversion you've got.
 
Those part numbers don't look familiar to me so I doubt I have them around, are they in the HM anywhere, in that case I might...

EDIT: I would be glad to buy the parts and build the circuit you want to test, unfortunately all of the good electronics supply houses in my area have gone out of business long ago. Is there a transistor that Radio Shack may carry that would work?
 
Last edited:
They're not used in the current HeaterMeter BOMs, but they are in the v4.2 boards. The board I'm mailing you you can try both ways and see if both work or neither work.

I tested the current draw on the servo signal line and both with and without the NPN current booster my servo draws about ~0.4mA on the signal line when it is high. I then swapped in a properly wired MOSFET driver with a 10k pullup and the drive voltage was... only 2.8V again! This had me scratching my head for a while until I realized that at 10k impedance a voltage droop of 2.2V would mean it was drawing 0.22mA. I swapped out the 10k pullup for a 1k and that lowered the droop to 0.25V or again 0.25mA. It seems pretty consistent that it draws somewhere between 200uA and 400uA which still puzzles me as to why the NPN transistor works at all-- the ATmega should drive way more current than is needed. I'm really trying to fully understand the magic so I can maybe use it in further development. I'm leaning toward the MOSFET voltage booster hopefully being just as good a solution for you on account of it seems more logical that it should be.

Whether that holds true in reality versus just theory would certainly be enlightening. Can you PM me an email address? I need to send you some pictures and instructions to go with the board.
 
Ralph, Fry's has NTE parts and they have a substitute for BS170 with NTE490 the 2N7000 can be substituted with NTE491. If Downers Grove to far, there is another electronics store Tri-state in Mount Prospect.
 
So my Mouser order was delivered unexpectedly early today and I have all the parts for the HMv4.2 build in hand, which means I have the MOSFET in stock to build and test the voltage booster circuit you were working on (Bryan). Is the circuit Peter F posted in this thread what you want me to use, or the circuit I see in the 4.2 schematic, or something different?
 
Last edited:
You must log in or register to reply here.

 

Back
Top