Pre modded servo boards for escs?

All things antweight

Moderators: petec, administrator, BeligerAnt

MySolderIsOlder
Posts: 69
Joined: Mon Jul 30, 2018 1:22 pm
Location: Kent
Contact:

Re: Pre modded servo boards for escs?

Post by MySolderIsOlder » Mon Apr 15, 2019 7:30 am

Having come quite late to the Antweight world, I missed out on the necessity for sourcing DIY ESCs made from servo boards. I was always curious about giving it a try but wasn't keen on the faff of balancing those tiny potentiometers. Hence, when I saw Mike's post about using the boards from digital continuous rotation servos, I got straight on to AliExpress and ordered up a 10-pack of those DomanRC units (15 quid, with postage). They arrived last week so got down to some testing over the weekend. Results were.... interesting.

For starters I tested them as servos - plugged into a Redcon 4-CH Rx, powered by a 6v NiMH battery. Four of the ten were DOA, the other six buzzed along in a sprightly manner, though I did notice an alarming tendency for some of them to creep slowly around, even when the stick on my DX6i was centered. This was true whichever channel I used and trimming didn't have any effect - nudging the adjuster-bar along, the slow creep would remain the same until a certain point, when normal rotation in the other direction kicked in. Should mention that I've never had this problem with any pukka ESC.

Notwithstanding this unpromising start, I proceeded to dismantle all ten units and extracted the tiny speed controllers for direct testing, both with an N20 and with a multimeter. As expected the four dead ones remained dead - no voltage at all on the motor leads. As for the other six, with the stick in neutral, they all gave a steady 0.3v, which was enough to make a zero-loaded motor idle at around 15rpm, though it only took a small amount of pressure to stop it moving. Adjusting the trims, it read a constant 0.3v until reaching the tipping point when it would jump to -1.5v
I'm sure I could cure this lack of a proper dead-zone with a small resistor or something. Slightly more alarming is what happens at the other end though - with the stick at maximum deflection, I never got more than 4v going to the motor. Switched the 6v NiMH battery for a fully charged 2S LiPO but same result. Thinking it might be the Receiver, I repeated the tests with an Orange Rx614 but again same results. Out of curiosity, I also dismantled an old Feetech FS90R (analogue C/R servo) that I had lying around from an Arduino kit. Didn't manage to get the potentiometer on this one any nearer a true 0v dead-zone either, although I did find that at full stick, this board was giving a more respectable 5.2v.

So... not entirely successful but an interesting experiment nonetheless. I will certainly find a use for these cheap and tiny ESCs (maybe time for my first cluster) but it's possible I was just unlucky in the usual gamble over Chinese manufacturing quality control. Game's not over yet though - while looking at the Doman servos on AliExpress I noticed that Feetech are also now making a digital C/R servo (the FT90R) so I've got a few of those on the slow boat from China and will post test-results when they eventually arrive.

Possible I'm expecting too much from servo boards - but given how cute and tiny they are, it would be a shame not to find a use for them (especially if I can find a suitable small PWM receiver to replace the currently unavailable Rx614).
Stuart (Anthony's dad)

User avatar
MarkR
Posts: 196
Joined: Mon Dec 18, 2017 12:46 pm
Location: Reading Hackspace
Contact:

Re: Pre modded servo boards for escs?

Post by MarkR » Mon Apr 15, 2019 7:53 am

Sounds more like a poor design or poor choices of parts, than a manufacturing defect.

A voltage drop of 2V at the kind of current we'd see through a N20 is extremely high, did you have access to an oscilloscope? Maybe the duty cycle is less than 100% even when the stick is at the extreme position?

--

Most of these parts we're working with (ESCs etc) don't have a clock crystal in, so they can't really measure 1500us accurately. So you're unlikely to get them dead centre, and it might vary with temperature, etc.

Radios, on the other hand, need an accurate crystal to work, so the digital radio should always send very close to 1500us when stick is centred.
Robot: Betsie - RaspberryPi controlled flipper bot with gyro stablisation - too clever for her own good?

Ant Ipodean
Posts: 33
Joined: Mon Apr 24, 2017 3:31 pm
Location: Auckland, New Zealand

Re: Pre modded servo boards for escs?

Post by Ant Ipodean » Tue Apr 16, 2019 10:47 am

Hi MySolderIsOlder
Not questioning your obviously thorough testing, but this is quite different to the results I have been getting.
I received another batch of 10 units recently for "stock", but haven't needed to dip into them yet.
You have me wondering now if the Chinese have been "tampering" with their original spec.
I am travelling at present, but will test them when I get home.
Mike.

MySolderIsOlder
Posts: 69
Joined: Mon Jul 30, 2018 1:22 pm
Location: Kent
Contact:

Re: Pre modded servo boards for escs?

Post by MySolderIsOlder » Tue Apr 16, 2019 12:37 pm

Hi Mike - please don't ever worry about questioning my testing - my understanding of electronics is rudimentary at best :D

I did wonder whether my el-cheapo Neoteck DMM was correctly reading RMS voltage, so I tried the same test with one of my few surviving Das-Micro ESCs. There however I was definitely seeing the full range of 0V (give or take ~20mV) up to 6V, so I think the test methodology is legit.

The real issue may well be changing spec. When I first tried putting in an order, it was subsequently cancelled due to 'stocking issues'. Then when I tried again a few days later, the Doman servos were back in stock and I was able to re-order, so maybe they were in the process of switching component suppliers? Perennial problem with ordering from China - you never really know who made the item, whatever the name badge may say, nor whether it will be the same when you next order one. I'm hoping I just got unlucky, as the ability to cobble together 2-channel ESCs for less than a fiver each is too tempting.

Do please let us know how you get on with your next batch.
Stuart (Anthony's dad)

Ant
Posts: 5
Joined: Tue Dec 13, 2016 10:51 am

Re: Pre modded servo boards for escs?

Post by Ant » Tue Apr 23, 2019 10:55 pm

Hi
We have tried modifying various servo boards with varying degrees of success.
Hobbyking hxt900 have to fit unbalanced resistors but work well although the trims move off as they get warm
Have tried some £1 servos from eBay but it is not possible to get them to stop- I assume the dead and is too small. In fact, when built up as a servo they seem to rely on the gear friction to hold stationary.
Hobbyking standard grey servos are much easier to solder and my son has built them into a 4wd pusher with each driving two motors.
After reading this thread and the hope of reducing solder burns(!) we ordered 10 of the doman continuous servos. Have tested them all in their supplier state and 9 are ok- all have the same trim offset to remain stationary. The last one has a constant slow rotation that when you try trimming out swaps to the other direction.
Don’t know how long they will work for.....

MySolderIsOlder
Posts: 69
Joined: Mon Jul 30, 2018 1:22 pm
Location: Kent
Contact:

Re: Pre modded servo boards for escs?

Post by MySolderIsOlder » Wed May 01, 2019 10:57 am

Quick update...
As mentioned in my earlier post, at the same time as ordering the Dorman servos, I also ordered a batch of Feetech FT90R c/r servos for comparison. They cost twice as much as the Dormans (£30 for 10) but I figured with Feetech being an established brand, they should be worth a try. Unlike Feetech's older FS90R servo, these use digital control boards, so again in theory no messing about with resistors to find the dead band. Took six weeks on the Silk Road but they arrived yesterday and have just done some quick tests with one of them. Initial results are extremely promising.

Image

First obvious difference from the Dormans is that pcb is slightly bigger (10x20x3.5 mm) and the motor lugs are soldered directly to it. De-soldering was pretty simple though and you're left with nice big holes for soldering motor leads - and a big round hole where the end of the motor used to be.

Image
Image


As before, tested by connecting to a Redcon 4-ch Rx & 6v NiMH and clipping a digital mulitmeter to the motor pads. This time however the results were perfect - clean 0v with the stick in neutral and +/- 6v at either end. Also tried briefly with a 2S LiPO and was reading +/- 8.2v to the motor, same as the battery voltage (the Redcon Rx is rated to 10V so need for a separate BEC). No sign of any drifting off the dead band, though I was only testing for a few minutes.

So... only tested one unit so far but will try to find time over the weekend for more comprehensive/destructive testing and do a proper write up.

Cheers,
Stuart
Stuart (Anthony's dad)

Ant Ipodean
Posts: 33
Joined: Mon Apr 24, 2017 3:31 pm
Location: Auckland, New Zealand

Re: Pre modded servo boards for escs?

Post by Ant Ipodean » Wed May 01, 2019 12:49 pm

I am back home now, and have had a chance to test some of the Doman 360 servos and ESCs ..

First I tested the 10 most recent ones (Ordered 17 Mar, Rcd 28 Mar)
I ran each one from Ch 1 of an FS-iA6 Rx with 4.92 volts in, and an FS-i6 Tx.
They all seem to do "what it says on the packet".

Next I ran them from a digital servo tester (with a range of 800 - 2200)
The dead zones all ranged between 1430-1450 and 1530-1550, so very good, although on average slightly more dead zone below 1500 than above.
I did not disassemble any of these, but rather tested some previous ones that I have in various setups.

Next I tested one in one of my Ants, driving an N20 with 7.46 volts input. (My lipos are all in storage mode)
On full throttle the maximum at the motor was 6.19 - 6.22 volts.
Interestingly, when I disconnected the motor and measured the ESC output on full throttle, it ONLY reads around 3.4 volts !
Presumably there is some kind of inter-reaction between the transistors and the motor loading to make it all work, but a full understanding of how the electrons are jumping around is a bit beyond my pay grade.

I also have a pair in a 4WD bot (4 x N20s) but setup to send the signals to a "2 DC Motor Drive Module Reversing PWM Speed Dual H Bridge L298N" board.
This one is running on 8 volts regulated (actual 7.9 volts), and has been driving flawlessly for a long time.

Then, modelled on this, I tried setting a pair up to feed signals to another "dual H bridge motor drive module" with higher voltage and power capabilities, with the idea of using it to drive Beetleweights.
This is just on the test bench, with the control circuits on 5 volts and the power side on 12 volts with 25mm gearmotors.
Whilst the motors respond correctly to the Tx sticks, they are just not getting enough volts, and there is a big voltage drop at the output of the little 360 ESCs.
So i presume this is something to do with backloading in the circuit, similar to when I measured the voltage output without a motor connected.
Or perhaps it is to do with a narrower or wider frequency range?

In another context, I have previously used these little servo ESCs to switch a micro relay for on-off switching of higher loads.
I found that they were much more stable in this application when I put a capacitor across their output leads.

To summarise ..
1. 4 out of 10 being DOA is very disappointing.
2. Maybe we are overthinking this - Just put a pair in a robot with 2S and 2 x N20s, and try driving it. I have done this with 4 robots now that ran fine.

Mike.

Ant Ipodean
Posts: 33
Joined: Mon Apr 24, 2017 3:31 pm
Location: Auckland, New Zealand

Re: Pre modded servo boards for escs?

Post by Ant Ipodean » Wed May 01, 2019 1:00 pm

Hi Stuart

I will be very interested to see how these new Feetech servo ESCs work out?

Mike.

MySolderIsOlder
Posts: 69
Joined: Mon Jul 30, 2018 1:22 pm
Location: Kent
Contact:

Test results

Post by MySolderIsOlder » Wed May 15, 2019 9:02 am

Regarding the differing experiences when testing the Doman servo boards, thanks to Mike (Ant Ipodean) the mystery is solved. He very kindly airmailed me one of his stock of Doman boards to test.

When I repeated my tests using Mike's board, my observations were the same as his, i.e. there is a clear 0v dead-band in the middle, at full stick the voltage is 2-3v below supply voltage when connected with just a multimeter but only around 1v below supply when the voltmeter is connected in parallel with an N20 motor. By comparison, my version of the Doman boards had around 0.3v (enough to turn a motor) in neutral and were 2-3v below supply regardless whether a motor was connected or not.
(I'm guessing that on both cases, the voltage drop is because the PWM isn;t reaching 100% duty cycle but with no access to an oscilloscope I can't confirm that. By comparison, the Feetech boards I've been using deliver the full battery voltage to the motors.)

Clearly, these differences are not accountable for by normal manufacturing quality variation - and make a huge difference to whether these servo boards are suitable for ESCs or not. However, it was immediately obvious on inspecting the boards that Doman must have changed design and/or suppliers at some point - the PCBs are not even the same colour and have a number of track differences. In terms of the SMD components, always hard to get definite IDs but...

Mike's Doman board (green PCB)
Regulator: V1QG
Driver: FZMOS-5X
Other: 2x resistors, 3x capacitors, unmarked 8-pin microcontroller

My Doman boards (blue PCB)
Regulator: 662x
Driver: FZMOS-2
Other: 1x resistor, 3x capacitors, unmarked 8-pin microcontroller

Spot the differences in these front and back views; my board is the blue one on the left in each case:
Image
Image



So.... hard to say whether it's worth buying the Doman servos for use as ESCs as it all depends which version they send you. Chances are, if you get the green board you'll be :lol: but if you get a blue board, you'll be :( No idea whether the blue boards were a permanent change or maybe just clearing older, faulty stock during temporary supply difficulties - but I'd be v interested to hear other peoples' experiences.

Personally, I'll stick with the Feetech FT90R boards (see other thread "Follow up on Digital CR Servo Board ESCs" for details). They are twice the price (£3-4 each from AliExpress, depending on quantity) and slightly longer - but being a well established brand, they should be more consistent and in all my testing so far they've been impeccable; no DOA units, clear dead band, full supply voltage to the motor and excellent proportional response.

Thanks again to the Auckland Antmeister for his assistance in clearing this up - and to Hogi for setting me on the 'pre-modified continuous rotation servos' path in the first place!

Stuart
Stuart (Anthony's dad)

MySolderIsOlder
Posts: 69
Joined: Mon Jul 30, 2018 1:22 pm
Location: Kent
Contact:

stop press

Post by MySolderIsOlder » Wed May 15, 2019 9:58 am

Doh!

Just after I hit submit on that last update, postman arrived bringing a second batch of 10x Feetech FT90R servos, with which I was planning to build up my stock of "Poor Man's Nano" controllers. And guess what? Yup - totally different boards from the last batch. This time, instead of the longer boards with the motor soldered on directly, the 20-pin quad-flat package micro and the AM1096 motor driver, these new ones have smaller (9x12mm) boards with short leads to the motor, an unmarked 8-pin DIP micro and TC118S motor drivers... :roll:

So much for my theory about established brands being more consistent :oops:

Cue another round of testing!
Stuart (Anthony's dad)

User avatar
LimaHotel
Posts: 151
Joined: Fri Mar 16, 2018 10:30 am
Location: W. Devon

Re: Pre modded servo boards for escs?

Post by LimaHotel » Wed May 15, 2019 10:11 am

I just had a paranoid look at the Doman boards I'd already ordered - Phew, mine are green!
I haven't had a chance to test them yet, but here's hoping!
Jiggy Sawdust (Grabber)
Jammin' Good with Weird and Flippy (Rear-hinged Flipper)
The Pushbot from Mars (4wd Pushbot)

Key Grip (Rear-hinged Flipper, currently out of service)

MySolderIsOlder
Posts: 69
Joined: Mon Jul 30, 2018 1:22 pm
Location: Kent
Contact:

Re: Pre modded servo boards for escs?

Post by MySolderIsOlder » Wed May 15, 2019 7:43 pm

So finally...
just tested the other type of Feetech FT90R esc as arrived today; what I’ll call the “Feetech short board”.
Good news is it’s got a clean dead band and supplies full voltage to the motor. Bad news is that the proportional response sucks - not much gradation between off and full throttle.
So, better than Doman blue boards, roughly on par with Doman green board, not a patch on Feetech “long board”.
Here are the front and back views
Image


Have ordered up a cheap servo tester so should be able to give more accurate views on proportional subtlety soon
Stuart (Anthony's dad)

User avatar
LimaHotel
Posts: 151
Joined: Fri Mar 16, 2018 10:30 am
Location: W. Devon

Re: Pre modded servo boards for escs?

Post by LimaHotel » Wed May 15, 2019 9:43 pm

Can you not compensate for the lack of proportionate control by adjusting the scaling on your tx? Or is it too bad for that?
Jiggy Sawdust (Grabber)
Jammin' Good with Weird and Flippy (Rear-hinged Flipper)
The Pushbot from Mars (4wd Pushbot)

Key Grip (Rear-hinged Flipper, currently out of service)

MySolderIsOlder
Posts: 69
Joined: Mon Jul 30, 2018 1:22 pm
Location: Kent
Contact:

and another thing...

Post by MySolderIsOlder » Mon May 20, 2019 10:08 am

OK, promise this will be my last update on this thread (if anyone's still reading)

Treated myself to an el-cheapo servo tester (HJ Digital Servo Tester - 7 quid from Amazon) to see if I could get a bit clearer about the discrepancies I was observing between these different servo boards. Results were most enlightening.

Dodgy testing in progress (here with the 'short-board' Feetech)...
Image


Firstly, I think I now know why I was seeing a much reduced voltage at the motors using the DomanRC boards. Standard DSM Tx/Rx combo is meant to give a control pulse in the 1-2ms range, with 1.5ms as zero - and I think that's pretty much what I get from my Spektrum DX6i with the default -100% to 100% endpoints. That was giving me +/- 4v with the DomanRC c/r servos. With the Digital Servo tester, which runs from 0.8-2.2ms, I was getting nearer 5.5v at the ends, and I was able to get something close to that after setting the Tx endpoints to 125%. In other words, looks like these DomanRC servos are designed to work over a wider than usual range of pulse widths. The description for one of their other digital servos (a 270 degree version) specifies a PWM range of 0.5-2.5ms is required for maximum movement - so I'm guessing with the 360 degree version you'd need a similar PWM range to get maximum voltage (100% duty cycle) to the motors.
Secondly, using the 'green-pcb' Doman board that Antipodean sent me, I got a pretty good proportional response over the full range. Dialing the tester up from 0.8ms to 1.4ms, voltage changed in a reasonably linear fashion from -5.5 to -2.3, then dropped suddenly into the dead band, kicking back in again to +2.3v at 1.6ms. So, a small cliff-edge either side of the dead band but still very good.

Doing the same test on the two variants of the Feetech FT90R boards, the long board showed a healthy degree of proportionality - steady -6.2v (full battery voltage) from 0.8 to 1.3ms, then a very smooth slope all the way down to zero at 1.45ms, picking up again at 1.55ms. This matches my experience of driving my test bot utilising these boards, where I was able to make it crawl along with a gentle stick movement but pushing the stick beyond about half way resulted in full speed. Might even be worth reducing the endpoints in this instance to increase the proportional stick range.
By contrast, the FT90R short board is about as proportional as our FPTP electoral system - it was delivering full voltage from 0.8 to 1.45ms then dropped to zero almost immediately, jumping back up to full voltage again at 1.55ms. Basically, it's on or it's off with a difference of maybe 30 microseconds in pulse width between the two extremes (it was quite hard to measure but the 'proportional phase' only seemed to be around 1.44-1.47ms so I don't think any amount of Tx tuning is going to help there.)

So... to conclude (phew!)... either the Feetech FT90R long board variant or the DomanRC 'green pcb' variant should make a fine ESC, though with the latter you'll want to ensure your Tx/Rx can go well beyond the 1.0-2.0ms pulse width range in order to get maximum speed. By contrast, the Doman blue PCB (no dead band) or FT90R short-board (no proportionality) versions should be avoided. Trouble is of course, you won't know which variant you've ordered until they arrive!

Promise I won't say any more about this, though I will be interested to hear other peoples' experiences of these servo boards.
Stuart
Stuart (Anthony's dad)

User avatar
peterwaller
Posts: 3678
Joined: Fri Feb 15, 2002 12:00 am
Location: Aylesbury Bucks
Contact:

Re: Pre modded servo boards for escs?

Post by peterwaller » Mon May 20, 2019 10:56 am

You don't need to apologize it is all good information you have been posting.
If people aren't interested they don't have to read it but I am sure it will be useful to those contemplating using servo boards. :D

Post Reply