Fullscreen HTML5Med Res (??MB)HTML4

I REMOTE CONTROL BY BOB YOUNG Switching frequencies in speed controllers - which is optimum? This month, we will be taking a close look at the topic of switching frequency and how it affects the design and operation of speed controllers. Some designers use 50Hz as the switching frequency while others prefer 2.5kHz As noted in an earlier column, it is most important to switch the FETs in the fractional throttle range to avoid overheating them and wasting undue power. What we have yet to discuss is the question of what frequency do you run the chopper at? Here we have a philosophical argument of the utmost subtlety and with far reaching consequences ifwe are to believe the proponents of the 2.5kHz school. Much ink has been spilled by the electric modelling fraternity arguing over the merits of 50Hz, 2.5kHz, quency we run at. If the FETs and motor are on 50% of the time and off for 50% of the time, the current consumption and thus FET and motor heating · should be a constant. The same applies for all other pulse widths and by definition this must be so, for this is exactly how the system works. A 50-50 duty cycle gives half power, 75-25 three quarter throttle and so on, regardless of the chopper frequency. Add to this the fact that most modellers have a desperate need for speed and jam the throttle wide open for "The basic argument runs that the 2.SkHz switching rate is more efficient than 50Hz switching in that the motor runs cooler. Also, at very small pulse widths (low throttle), the control is much smoother and more precise". or frequencies in between. Some stalwarts have even changed sides and renounced their earlier views and have thus added additional confusion to an already perplexing argument. 50:50 duty cycle So what is the argument all about? At first glance it appears a storm in a teacup for a 50-50 duty cycle is a 5050 duty cycle regardless of what fre- most of the motor run anyway. At this point, the controller moves out of pulse (or switch) mode into straight DC and one really must wonder just what the fuss is all about when, for about 90% of the time, there is no pulsing in the system at all. The motor is running flat out. The basic argument runs that the 2.5kHz switching rate is more efficient than the 50Hz switching in that the motor runs cooler and is therefore subject to less demagnetising from heat. Also, at very small pulse widths (low throttle), the control is much smoother and more precise. There is also some talk of the 50Hz pulsing being more destructive to the magnets than the higher frequencies. 50Hz advantages Against this, the proponents of 50Hz systems claim quite rightly a lower component count and therefore higher reliability, smaller size and weight, and lower cost. The lower component count derives from the fact that the 50Hz is generated by th~ Rx decoder, whereas in the 2.5kHz system the 50Hz must be converted into 2.5kHz by a separate 2.5kHz oscillator. It is safe to say that it is the choice between these two fundamentals that shapes the basic design of any speed controller and the argument rages on, still unresolved. So who is right? That is what I have been trying to establish for the past three months. I must point out here that the problem is a lot more complicated than you might think and I can understand why it has never been fully resolved. To begin with, the motor is an inductive load when running and thus subject to the effects of frequency on impedance. This is compounded by the fact that the armature is switching at a rate related to the RPM and the number of poles in the commutator. The net result is a complex network of switching transients, back EMF transients and spark generated noise, all of which are changing in relation to one another as the chopper pulse width modulation and motor RPM vary. APRIL 1992 53 This means that the analysis is well out of the domain of the average electronics buff as it requires some quite specialised test equipment. For my tests, I set myself up with some quite basic equipment and it was not until I attempted to analyse the very surprising results that I realized how difficult a full analysis would be if the job was to be done correctly. My initial test set-up involved a heavy duty battery (on float charge) , a pulse generator which could be varied over the full range of wanted frequencies, a tachometer, a moving coil ·ammeter to monitor the current, and a storage oscilloscope to monitor the various voltage and current waveforms. The pulse generator was checked carefully for pulse width against frequency and gave a consistent 52-48 duty cycle over the usable range of the FETs. The tachometer was a photocell type. Two IRFZ44 FETs were used without base stopping resistors to drive a Leisure 05 stock motor and an 8 x 4 propeller (direct drive). Straight DC drive current was 28 amps, at 10,500 RPM. You can now afford a sate II ite TV system For many years you have probably looked at satellite TV systems and thought "one day". You can now purchase the following K-band system for only: $995.00 Here's what you get: * A 1.8-metre prime focus dish antenna, complete with all the mounting hardware. * One super low-noise LNB (1.4dB or better). * One Ku-band feedhorn and a magnetic signal polariser. * * lnfrared remote control satellite receiver with selectable IF audio 30 metres of low-loss coaxial cable with a single pair control line. & bandwidth, polarity & digital readout. Your receiver is pre-programmed to the popular AUSSAT transponders via the internal EEPROM memory. This unit is also suitable for C-band applications. Test results Table 1 gives the test results. As you can see, the revs and current drain are reduced as the frequency is raised, being a minimum at a pulse switching frequency of lkHz. Above that frequency, the revs rise but the current stays lower than at 50Hz. I am at a loss to present a definite solution to the curious results shown. Call, fax or write to: AV-COMM PTY LTD PO BOX 386, NORTHBRIDGE NSW 2063. Phone (02) 949 7417 Fax (02) 949 7095 All items are available separately. Ask about our C-band LNBs, NTSC 0 to-PAL converters, video time date generators, FM 2 & EPAL & Pay TV hardware. I I II ----------YES GARRY, please send me more information on K-band satellite systems. Name .............................................. I I II .___________ . 54 Phone .............................................. ACN 002 174 478 SILICON CHIP 01/92 Frequency RPM Current (A) 50Hz 500Hz 1kHz 1.5kHz 2kHz 2.5kHz 3kHz 5kHz 7600 7500 6800 6900 7200 7400 7500 7700 16.5 12.0 7.0 7.3 7.8 8.0 8.5 9.0 the 5kHz point but it must be remembered that I was only using two FETs. Six FETs will provide a much greater input capacity which will cause problems at the higher chopper frequencies. There is little doubt that the efficiency improves with frequency. Reference to Table 1 shows the current at 2.5kHz is approximately half that at the 50Hz figure for virtually the same RPM. On the other hand, I am not sure that the meter reading is a true indication of the current drawn. It was similarly difficult to interpret the current waveforms taken at various frequencies and I will need to take more definitive measurements before I can be sure of the relative merits of switching at 2.5kHz. Temperature measurements In view of the doubts about the current meter, temperature measurements taken after four minutes of run- "So there you have it - just as the argument for 2.SkHz switching predicted. It gives greater efficiency, cooler running and smoother control. Just don't ask me to explain it". I Address........................................... I I I I ........................... P/code ................ I II I TABLE 1 II I Although I did not think it important at the time to record the battery terminal voltage, I did note it mentally and the higher currents were associated with a lower terminal voltage, longer run times and longer charge times between runs. I also made a measurement at lOkHz but the reading was to my mind suspect, in that FET gate capacity was starting to distort the input. There was however no sign of distortion at ning gave a case temperature of 51.4°C for 50Hz operation and 37.7°C for 2. 5kHz operation. These case temperatures (maximum) were measured after the cooling effect of the propeller wash stopped and the internal heat had soaked through to the case. So there you have it - just as the argument for 2.5kHz switching predicted. It gives greater efficiency, cooler running and smoother control. Just don't ask me to explain it. But in spite of the above results, I find myself leaning very heavily towards the concept of a 50Hz controller. The resulting controller will be simpler and much less expensive than a 2.5kHz design. In my experience, these are very important points and as I have pointed out above, the controller will spend most of its time flat out anyhow. Other approaches Now let us discuss the ways other designers have approached the problem. The first example is a simple 2.5kHz controller with no braking. This controller is very smooth and quite linear in operation. It has six FETs which provide ample current for most applications. A voltage tripler provides 12.5V at the gates from the 4.8V Rx battery. It is a very nice little controller. I also have a circuit of European origin using the least components I have ever seen in any controller. One wonders how well it works. This is an opto-coupled unit to minimize noise fed back into the Rx from the motor drive circuit. It is fitted with a backEMF brake (dynamic braking) and again one wonders just how well that brake circuit works. From bitter experience, I have learned that the ON resistance of the transistor across the motor must be less than 100 milliohms for any braking effect to be achieved, which means that it must be driven hard. It has no voltage tripler and the drive voltage for both the forward and braking FETs is derived from the motor battery which is in this case quite adequate, being in the range of 10-35V. The disadvantage is that as the motor volts fall, so do the drive and braking voltages. Noise is also a bigger problem as the motor battery is coupled into the drive electronics and so an optocoupler is almost mandatory. It was obviously designed with model aircraft usage in mind, as a 7.2Vbattery would not provide sufficient drive to turn "From bitter experience, I have learned that the on resistance of the transistor across the motor must be less than 100 milliohms for any braking effect to be achieved, which means that it must be driven hard". the FETs hard on. It is typically European in approach, showing concern over feedback noise but unusual in using 50Hz. Another circuit uses 50Hz operation and has several clever features, including braking. Separate decoders drive the forward and braking FETs so that the brake cannot come on whilst forward is energized and vice versa. If this did happen, it would provide a dead short through the braking and forward FETs and destroy the controller. The circuit also has a voltage tripler which provides heaps of drive to both sets of FETs. This unit has been designed specifically for cars and uses a battery eliminator. The problem with battery eliminators is that the Rx runs off the motor drive batteries which eventu- Yokogawa DL1100 Oscilloscope - continuedfromp.16 press the "Initialize" button. This brings up an "Initial Exec" message on the screen, prompting you to press one of the softkeys (by the way, they're called "softkeys" because their function changes with each new screen menu). You might wonder why you have to press two keys to initialise the scope when it would be easier to press one. The same comment could go for the Auto Setup routine . And for that matter, you might ask why the machine could not initialise itself automatically at switch on. The scope could undoubtedly have ally go flat and thus all control is lost - not good in an aircraft. This type of Rx supply must also be filtered very carefully if motor noise is to be kept out of the circuit. There are also reversing controllers but these have a fundamental problem. The drive motor is included in a bridge circuit (similar to the Rail power controller featured in this month's issue) and thus there is double the volt- been made to automatically initialise itself at switch on but then there would not have been the convenience of having the last used settings saved. And the idea of making you press a soft key after pressing a front panel button stops you from accidentally wiping out existing settings . If you do press the wrong button and it brings up a screen menu that you don't want, all you do is press "Menu Off" and that clears it. Pressing it again brings the last menu back. From the foregoing it should be clear that the Yokogawa DLl 100 2 channel 100MHz digital oscilloscope is a age drop across the FETs as there is always one set of FETs on either side of the motor. For this reason, reversing controllers are not popular with the speed fraternity. They are, however, a must where total control over the model is called for. The final design Note that none of these circuits has all of the features considered desirable by the modern modelling fraternity so there is plenty of scope for new designs. Drawing from the above , our proposed design is a now a little firmer in that it will use 50Hz switching, dynamic braking, drive electronics working from the Rx battery, a free-running voltage tripler and, as a result of this battery isolation, no optocouplers. SC highly flexible and powerful instrument. It takes some time to become familiar with all its features and use them to the fullest. We had only a few days with it but in that time we have been very impressed. It is a fine instrument. The DL 1100 is priced at $4900 which includes the GP-IB interface, while the optional built-in thermal printer is an additional $750, as is the RS232 interface. These prices do not include sales tax. For further information, contact Tony Richardson at Yokogawa Australia Pty Ltd, Centrecourt D3, 25-27 Paul Street North, North Ryde, NSW 21_13. Phone (02) 805 0699. SC APRIL 1992 55