Thread Tools
Aug 27, 2006, 12:00 PM
Registered User
pilotpete2's Avatar
Adding 9000uF of filtering (9 x 1500uF Low-ESR caps) improves the situation dramatically. Voltage ripple is now very low, less than 50mV (I didn't bother to photograph it). At full throttle the current ripple is less than 1A, and at 2/3rd throttle it's about 1.5A with a much smoother waveform.[/QUOTE]

Thank you for posting those O' Scope shots, obviously I was off base in thinking that the input caps would provide enough filtering so that a Whatt-meter would be seeing essentialy pure DC, Not
Sign up now
to remove ads between posts
Aug 27, 2006, 10:14 PM
Got shenpa?
flieslikeabeagle's Avatar
Bruce, yes, thanks for the oscilloscope photos, the amount and spikiness of the current waveform (before you added the filter caps) was an eye-opener, for sure.

Aug 28, 2006, 11:29 AM
Dieselized User
gkamysz's Avatar
Interesting. What is the switching frequency of that ESC? (just did the math, ~7kHz) I wonder how the input current and voltage waveform compares to a battery source. I know a couple years back in Taiwan and China modelers were adding capacitors to LiPo packs and significatly reducing temperatures of packs during use.

Aug 28, 2006, 11:44 AM
jrb's Avatar
IIRC a year or so ago Patrick (CC) mentioned that sine was ESC were on the way!

That would probably help on the motor side of things. The battery side? More losses in the ESC?

Variable speed drives are rapidly winning favor in the industrial world; a lot of our pumps use them these days since its flow is directly proportional to rpm.

I wonder if the industrial ESCs are sine wave or just like ours? Or, maybe both – you get what you pay for.
Aug 28, 2006, 01:19 PM
Registered User
hul's Avatar
Originally Posted by jrb
IIRC a year or so ago Patrick (CC) mentioned that sine was ESC were on the way!
these people have it already:
but it's not cheap, 42V/110A for 700 Euros or 220A for 800

Aug 28, 2006, 06:15 PM
Registered User
Excellent input from Bruce. Helps to illustrate the hurried comment I had made: that a RMS voltage meter reading multipled with a RMS current reading does NOT produce a valid result for power. The common approach would be to sample V & I, then multiple point-by-point, etc. It could be done in the analog domain, but probably only a handful of people remember how to do that.
As to the "continuous" nature of voltage, current, time, etc., certain physicists, cosmologists, & DSP users would have a different viewpoint.
Aug 28, 2006, 06:40 PM
Got shenpa?
flieslikeabeagle's Avatar
Originally Posted by moyg
As to the "continuous" nature of voltage, current, time, etc., certain physicists, cosmologists, & DSP users would have a different viewpoint.
Well, current is quantized at the level of one electron charge. Since there are about 6,242,000,000,000,000,000 electrons per second in a current of one ampere, we can safely ignore the very fine-grained quantization for the most part (unless you're focussing on shot noise, which arises because of that quantization).

DSP? That's a matter of sampling a continuous quantity, turning it into a discrete one. The sound pressure level in front of the microphone was a continuous quantity before the A/D converters in the preamp digitized it...unless, once again, you are going to worry about the very fine-grained quantization from the individual molecules in the air in the room.

Time? Time was still a continuous variable in all the quantum mechanics classes I took at college. I don't know what the QED and String theory folks are doing with time these days, but classical quantum mechanics (if I may coin the term) still treats time as a continuous variable.

Aug 31, 2006, 12:37 AM
Registered User
I took some measurements last night and tonight on a PHX25 in the lab.

I don't have an RPM meter that likes indoor lighting, so I wasn't able to record RPM. But instead I focused on 1) understanding MOSFET switching times and 2) comparing RMS readings of battery current with Astro watt meter readings.

I measured mosfet gate drive at about 9.4V and 2.2 uS rise and fall times. That's quite a bit slower than the hardware is capable of, so I'll re-check. If in fact correct, then at 14A that means 99% throttle has about 5.5W in switching losses while 100% has about 0.25W. 50% would have about 2.2W. Remember that switching losses are a function of how much current is flowing through the mosfet, so as throttle decreases the losses also decrease. The HW is capable of switching times closer to 100 nS.

Next I looked at battery current. Instead of a battery I used a 13V 25A supply. To measure current, I used a 0.01 ohm shunt married to a Maxim MAX4372 current sense amp. This gave gave a full scale around 15A or so. I calibrated using a constant current supply, and adjusted the math function on the scope to deliver 1A/V (required 1.83 multiplier). Linearity was checked from 14V to 6V and was just 5 mA over that range with 2A into shunt. Good to go.

The shunt was placed after the astro meter, the motor was a 1200KV Frio with an 8x8 prop.

The scope is a Tektronix TDS7054, and I picked 125Msps sample rate.

Quick summary: Astro meter readings are pretty dicey when not at full throttle. In the first plot (13.8V) note the scope is reporting 5.26Arms, and 4.3Aavg, while the Astro meter reported 2.3A. That's a substantial error, but unfortunately is an error that would make the efficiency appear better than it actually is so it doesn't explain the efficiency drop we've been seeing. Note in the first plot that the current ripple is about 12A, and the amount of ripple is seems key to confusing the astro meter. Note that you can clearly see the 11 KHz PWM frequency.

In the second plot, it's at 6.4V in and full throttle. Note the 11 KHz PWM is gone (since it's full throttle), and the interval you see is a commutation cycle. So as the voltage is increased, this period decreases. For this measurement, the astro meter reported 8.7A versus the scope's 8.66Arms and 8.63Aavg. Thus there was good agreement there.

In the last plot, it's at 6.7V and partial throttle. The astro meter reported 0.8A, the scope reported 1.16Arms and 0.903Aavg. Fairly close.

I'll measure some more hopefully tomorrow.
Oct 15, 2006, 04:55 PM
Always right
Hovertime's Avatar
Wow, lots of math used in this thread!
Are you guys ready to draw some conclusions and post a summary for the mere mortals?
Oct 15, 2006, 07:49 PM
Registered User
Before I retired from brushless drives in industrial machines I found that a lot of complete, expensive and brand name drives had some serious SORE spots in RPM and load combinations. If allowed to operate, the drives would go into temerature shutdowns. Motors and or the ESC could go to a shutdown.

I could program out the SORE spots with enough shut downs recorded.

We unkowningly continue to fly models at these SORE points and lose a motor or drive over time.

Not much we can do about those points.
Oct 15, 2006, 11:11 PM
Got more toys than my kid
macr0t0r's Avatar
Conclusion? Eh....we're not there yet.
Best I can tell, it is largely dependent on how fast and how intelligent the ESC is. On top of that, we aren't even sure if the current reading from a Wattmeter is accurate at partial throttle due to the pulses.

All I can say is, keep reading!

- Jim
Oct 16, 2006, 09:16 AM
Registered User
pilotpete2's Avatar
Maybe we need to add some big old filter caps to our Wattmeters to eliminate that situation, but then the surge on connection might be hard on the batteries and or meter unless we put a SPST switch in the battery side shunted by a resistor to allow the cap to charge slowly before closing the switch.

Were those brushless motors running with BLDC ESCs (closed loop/commutated) or were they run on VFDs (open loop) as AC synchronous motors?
Regards to all,
Oct 16, 2006, 10:14 PM
Registered User
Open loop had 99.99 % of these problems.

Some closed loops could be programed to actually sense the problem and would pick the closest good speed.
Feb 18, 2007, 10:58 AM
Registered User
Originally Posted by hul
these people have it already:
but it's not cheap, 42V/110A for 700 Euros or 220A for 800

Grueizi Hans,

thanx for mentioning our website! :-)

Maybe we can add some interesting measurment data to this discussion:

- input current (mean + trms)
- input power (effective + reflective)
- ´normal´ block kommutated ESC vs. SinusLeistungsSteller

You will find it here: (sorry: only german text available ...)


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Discussion Peak current at partial throttle Lnagel Power Systems 19 Oct 19, 2006 03:18 AM
Discussion less amps with more lipos for a given RPM at partial throttle? Nethole Batteries and Chargers 30 Oct 12, 2006 05:50 AM
Are outrunners more efficient at partial throttle DWS Power Systems 3 Feb 18, 2005 08:57 AM
Is flying at partial throttle with a sensorless brushless motor bad? Mitch G Power Systems 6 Dec 12, 2001 02:15 PM
ZAGI 400-X ; power cutting out to motor after flying at full throttle Joe-z Foamies (Kits) 29 Jul 15, 2001 10:51 PM