Fullscreen HTML5Med Res (??MB)HTML4

REMOTE CONTROL BY BOB YOUNG Building & using the low-cost speed controller; Pt.2 The low-cost speed controller presented last month is a surface mount design on a PC board measuring 55 x 38mm. It will be supplied with the surface mount components installed & the board tested. All the constructor has to do is fit the FETs and the servo input lead. The first step in the assembly is to fit a suitable servo lead to match the receiver you intend to use. These leads should be available from your local hobby shop. Modern servo leads usually come as a 200mm ribbon cable fitted with a 3-pin crimp connector. However, each brand ofR/C set seems to use a different connector and they do not follow any convention in lead colouring or lead position. J.R. and Futaba, for example, use Neg, Pos, Signal whereas Sanwa and K.O use Pos, Neg, Signal, so please take care here. Lead colouring is also messy. The usual convention has a dark colour or black for negative, a red \orange for positive and a neutral colour for signal, but don't be fooled. For example, Futaba use Black, Red and White which is logical but J.R. use Brown, Red and Orange (Sig) This version of the speed controller has eight Mosfets connected in parallel but you can use less, depending on the current requirements of the motor. 40 SILICON CHIP which is confusing. Sanwa receivers have Black with Red trace (Pos), Black (Neg) and Black (Sig) which is crazy. I did warn you to be careful! Fig.1 shows the component layout on the top of the PC board and also the location of the three input wiring pins. For those who feel this is all getting too much, fully assembled, tested and tuned units, with the lead fitted, will be available from Silvertone Electronics (see footnote). Those intrepid souls who can cope with this type of confusion should read on. Parallel Mosfets The PC board provides for up to eight power Mosfets connected in parallel to handle high motor currents. The basic concept of this project is to allow the experimenter to fit as many and whatever FETs he can afford, to suit his particular application. IRFZ30s are affordable, yet will cause only a 125mV loss at 10 amps with four in use. If very high current motors are to be · used, eight IRFZ40s will cut the voltage loss in half. Some of the more exotic FETs such as the IRFZ44, SMP60NO6-18, STVHD90 or the new National DMOS NDP 705, 706 range will give even better figures but at a much higher price. Table 1 lists some test results that were obtained using various numbers and types of FETs. The Speed lB controller has been tested with 30 cells at 46A but lack of suitable test motors has prechided testing to higher figures to date. The unit in this test was fitted with 8 x SMP60NO6 FETs (23mQ) and dropped 0.41 V. This voltage drop seems rather high considering that the eight paralleled FETs should have a combined "ON" resistance of 2. 9mQ. Where is all that extra voltage drop coming from? At this point, we must break off for a spot of theory on high current design. PCB and SCHEMATIC CAD I • • -- •:: • •~ m i' .TV. .IFwAMPLIFIE Keeping it cool Fig.3 shows the variation in drainsource resistance (Ros) versus temp era ture for a National NDP705 Mosfet. This illustrates the need for FETs to be kept as cool as possible. If the . temperature is allowed to rise, Ros will rise, increasing the power dissipation across the FETs, which will in turn increase Ros and leads to further heating. At some point, Ros will stabilise but, by this time, considerable power will be lost to the motor as heat losses in the FETs. I might add here that this factor allows FETs to be parallelled, so it is not all bad. In effect, one FET (lowest Ros) in the group will pass more current and warm its junction, thereby increasing Ros and reducing the current. This effect is not perfect in actual operation and some matching of the FETs is recommended for best results at very high currents. However, this problem has other ramifications which must be clearly understood if the full potential of these exotic semiconductors is to be realised. For example, the compartment housing the speed controller must be well ventilated or better still, the FET heatsink tabs should be fully exposed to the slipstream outside the model. Fig.1: here's how the components are mounted on the top of the Speed 1B PC board. The board will be supplied with the zener diode, trimpots & links fitted. All you will have to do is fit the Mosfets & the servo lead. 275C shows that a 2oz, 0.160-inch track (the width of the FET tracks in the Speedlb) will produce a 45°C temperature rise at about 20A. Even though the PC board design has the gate leads cranked outwards, the space between the FET source and drain leads does not allow any greater track width. Thus, I recommend that if you intend to run very EASY-PC • Runs on PC/XT/AT/286/386 with Hercules, CGA, EGA or VGA. • Design Single sided, Double sided and Multilayer boards • Provides Surface Mount support • Standard output includes Dot Matrix/Laser/Inkjet printers, Pen Plotters, Photo-plotters and NC Drill • Award winning EASY-PC is in use in over 12,000 installations in 70 Countries World-Wide Heating of copper tracks • Superbly Easy to use Heat generated in the FETs is only part of the· problem, however. In the unit tested to 46A, the tracks connecting the FETs were almost as hot as the FETs themselves. This was due to the fact that the prototype was constructed from "one ounce" (ie, one ounce per square foot or 35 microns thick) copper laminate and no additio_n al solder or bus wire was applied to these tracks. Thus, the track resistance was really too high for this ~urrent. That's where most of the extra voltage drop was coming from. The production boards for this project will be made from two-ounce copper laminate but that only goes part of the way to solving the problem. Reference to the US MIL STD • Not Copy Protected Options: • 1000 piece Schematic symbol library • Surface Mount symbol library • Gerber Import facility For full info 'phone, fax or write: BTC Fig.2: this diagram shows the surface mount components on the Speed 1B board. Note that some of the resistors are zero-ohm links & were not shown in the circuit published last month. The board comes with all surfacemount components fitted. PO BOX432 GARBUTT 4814 QLD. PH (077) 21 5299 FAX (077) 21 5930 DECEMBER 1992 41 This close-up view shows the surface mount components on the copper side of the board. Note the heavy solder coating applied to the FET tracks. high currents, lay down a very heavy coating of solder over these tracks after the FETs are soldered in place. The PC board supplied is tinned on these tracks , thus providing some additional conductor thickness and allowing this additional solder to be applied. The rest of the PC board is coated with a solder mask which prevents solder being applied to the tracks, but it will help to minimise corrosion in humid atmospheres such as boats. In theory, we should be able to run currents in excess of 30A and hold down the temperature rise to reasonable limits. However, I have effectively doubled the track width by bringing the power leads into the geometric centre of the tracks. This is a bit of a nuisance if a heatsink is to be fitted but is well worth the trouble. As each side carries only half of the current, the current handling capability of the Speed 1B PC board can be around 60A or more with some track reinforcement, without undue temperature rise. The important point in all of this is that the heat generated in the PC board tracks will be conducted up into the FET junctions, resulting in a further increase in Ros and even more losses, so keep those tracks well reinforced if you intend to use very high currents. Balancing currents Since the design of the PC board has the power leads brought to the centre of the FET tracks, they must be mounted symmetrically to achieve the current balance referred to above. So if you are using four FETs, they must be mounted with two on each side of 42 SILICON CHIP The PC board fits into a small plastic case, with the metal tabs of the FETs protruding through a hole in the lid to ensure adequate cooling. the power leads. Incidentally, this is a good rule to observe in any parallel FET PC board layout, as non symmetry in the PC board, particularly in the gate tracks, can cause incorrect load sharing. Having taken such care to mini1.8 ,i 1.6 J,"' ~ z: e u ~ ..."' 1.2 v.., Q ':I -c 2 "' i 0.8 / v" vV lo= '40A VGS = 10V 1· 0.6 -50 -25 0 I 25 50 75 100 125 150 175 TJ, JUNCTION TEMPERATURE ("C) Fig.3: this diagram, taken from data on the National Semiconductor NDP706A, shows the effect of temperature on Ros• As you can see, increasing the junction temperate from 25°C to 100°c increases the ON resistance by about 35% which means that the FETs should be kept as cool as possible. PLUG TO RECEIVER + ► 0:: SPEED CONTROL .......,< w IC 0 ,.... 0 BLACK GREEN Fig.4: suggested wiring arrangement for connecting the Speed 1B board to a typical radio controlled device. mise resistance losses in the speed controller, don't throw it all away in the connecting wiring. When I first began researching the speed controller project, I sought the advice of various modellers who I knew were considered experts in the field. I was regaled with stories of ZkW motors using 30 odd cells and drawing up to 70A. I also read manufacturers' specifications which stated that their speed controllers were good for 1000A or more. Now the question I must ask here is what kind of PC board and interwiring were these people using? We have already discussed the problems of track width so let's now look at the interwiring. Hartland Cables (Aust.) make an 801.25mm PVC coated cable, 4.7mm OD, which they rate at 40A. It is thicker than any cable that I have ever encountered on a commercial speed controller. Once again, if the temperature rises in the interwiring, this will find its way into the FET junctions and increase RosAll of these heat losses will rob the motor of much needed power and result in overall inefficiencies in the system. For currents up to 30A (most modelling applications), there are few problems. However, if you intend to push the system harder than this, the following recommendations are applicable: (1). Keep the FETs well ventilated. (2). If very high currents are required, lay a heavy coating of solder on the FET tracks or, better still, solder tinned copper wire or even solder wick along the tracks to reduce the resistance. (3). Use the best heavy duty cable Table 1 · FET Selection FUSE WIRE SOLDERED TO PCB >< Type Number Voltage Drop Current BUK455-60 8 100mV 25A STHDV90 4 110mV 25A STHDV90 8 60mV 30A SMP60NO6 8 410mV 46A PCB COPPER CONNECTING WIRE SOLDERED TO PCB Fig.5: suggested method for including a fuse in the wiring to the speed controller. Conventional inline fuses are too bulky and have high resistance losses. available for the interconnections and keep the.m short. Choice of FETs The more FETs used, for any given FET type, the lower the Source/Drain voltage drop and therefore the lower the heat dissipated in the FET junctions. The balance here is between cost and performance. Some of the more exotic FETs can cost up to $12, so you can be looking at $100 worth of FETs in an 8-FET unit. Alternatively, much more care can be put into the heatsinking to take advantage of cheaper FETs run at a lower temperature (and thus lower ON resistance). Thus, the IRFZ30s may be attached to small aluminium plates for heatsinking. The BUK455-60A is another useful and cheap FET. Note, however, that heatsinks are not necessary on systems with only moderate current drains, even with only a few FETs fitted. The flight battery can be anything from 4-30 cells. And although there has been no testing beyond 46A to date, I see no reason why the unit will not handle currents much higher than this. The interwiring presents the biggest problem to my mind. For the competition types who want to win at all costs, the sky is the limit. They should fit the very best FETs that money can buy: SMP6DNO6-18 (lBMQ), NDP705A or NDP706A (15MQ), STVDH90 (23MQ). They should also use as many FETs as possible and provide them with airflow cooling. To do this, cut a small hole through the fuselage, hull or car body and mount the circuit board so the FETs are outside, in the airstream. Take care that you don't static zap the Mosfets. Don't remove them from their anti-static bags until you're ready to install them and connect the tip of your soldering iron to the ground track on the PC board (via a clip lead) when soldering their leads. Use heavy gauge wire to connect the speed control to the motor and battery as discussed earlier. The best wire is the multi-strand instrument type cable. It has a very low resistance per foot and is very soft and flexible. The negative battery lead goes to the Mosfet terminals marked MZ , the positive battery lead goes to the positive motor terminal, and the negative motor lead goes to the Mosfet terminals marked "Ml" (Drain) - see Fig.4. Please, please use a fuse in one of the flight battery leads. They may add weight and take up space but this is Table 2 - Fuse Selection B &$Gauge Fusing Current 30 10 28 14.5 27 17.4 26 20.5 24 29.2 preferable to a burned out motor or speed control. The wiring diagram of Fig.4 shows the fuse in the positive lead but anywhere in series will do. There is no track fuse option on SpeedlB and I would strongly suggest that a small PC board fuse, as shown in Fig.5, be constructed and fitted in place of the standard type fuseholder (the latter are bulky and have a higher than desirable resistance). A fuse in which all connections are soldered gives the lowest resistance, even if it is not quite as convenient to use in the field. Table 2 shows the fusing current of different B&S gauges of tinned copper wire. For example, 20-gauge wire has a fusing current of about 58A. It would be a good choice if your motor draws around 35A at the chosen flight battery voltage. Wrap the fuse in fibreglass tape to avoid problems with molten copper splattering everywhere if it blows. Don't forget the diode, capacitor and varistor across the motor terminals. These must be mounted directly across the motor terminals with the shortest leads possible. Yes, I realise that soldering three leads to each terminal is a real pain and that modern commercial speed controllers have the diode built into the speed control module for the customer's convenience. I also realise that by not putting the diode in the unit, the user may be tempted to leave it out. Don't do that. - it will spell disaster for the Mosfets. Next month, I will describe the installation procedure and tell you how to adjust the throttle direction and sensitivity. SC 22 41.2 21 49.3 Where to buy the kit 20 58.6 19 69.7 18 82.4 17 98.6 16 Footnote: a kit of parts for this speed controller will be available from Silvertone Electronics. Phone Bob Young on (02) 533 3517 for pricing details. 117 DECEMBER 1992 43