The choice of power supply to enable the amplifier to achieve it's design performance means that in order to cope with all mains supply variations , transformer regulation etc a REGULATED supply must be used .This means that some form of control element is used to generate a steady supply voltage irrespective of the power being drawn by the active device in the power amplifier .
There are two type of power supply design which were evaluated for this project , conventional 50Hz & switch mode , each has it's own advantages & disadvantages. In the end after due consideration of the facts it was decided to use a standard 50Hz type of transformer to provide the relevant supply isolation & reduction . The main power losses are as follows , Bridge rectifier 2 x Vf x If this is common to all power supplies so with 28 volt output required this equates to 2 x 1v3 x 20 for a 400Watt power supply ( 52 watts ) However this can be reduced by the adoption of the higher supply voltage ( 50v ) reducing the current rating to produce the 400Watt ( 2 x 1v3 x 10 ) 26 watts .The control element of the linear supply is required to dissipate the " surplus " power this means that to cope with the mains supply & transformer regulation allowing for a 10% regulation limit on the transformer in conjunction with + 6 / -10% on mains supply limits means that for a 28 Volt supply ,a 30 volts AC transformer would be required ( @ 17 A ).During a mains supply minimum this gives 27 volts AC when rectified this gives 38 volts allowing for 1volt in the current sensing & emitter balancing resistors means that 130 watts must be dissipated , this rises to 350 watts at full output seriously reducing efficiency , even the use of a higher supply voltage for the amplifier only marginally improves the situation . ( Using a 35 volt AC transformer to give a nominal 40 - 45 volt supply ). So using the higher supply voltage for the amplifier to reduce the current requirements will in turn reduce the thermal requirements of the power supply's Heatsink .
If instead of using a linear regulator we use a switching regulator of the " Buck " type the losses involved are down to If x Vf in the series switch & If x Vf for the commutating diode , by using suitable choices of components these losses become negligible compared to those of the " linear " regulator. Typical losses in the series switch are 10 x 10 x 0.1 ( I2R DS) 10W .
From the application note " Designing of Switched Mode Power Supplies " by Motorola the commutating diode should be chosen for as low a possible Vf @ 1.5 x Iout . Use of a " Schottky " or Ultra fast rectifier is mandatory in view of the use of the switching frequency used , (typically 100KHz ) preference being given to the Schottky type due to it's low Vf this does however limit the choice of supply voltage out to 50 V max . Typically .65 - .85 volts @ 15A ( 12 Watts max) Ultra fast rectifiers have a Vf of upto 1.5 volts meaning upto 25 watts . This means that the regulator heatsink only requires to dissipate around 25 - 50 watts as opposed to between 130 - 350 watts
The power dissipation at the maximum mains supply voltage limit is actually reduced in the "Buck " regulator compared to " normal " supply values since the power switch is actually conducting for a lower duty cycle therefore producing less heat ! The choice of using a secondary switching regulator with a standard 50Hz mains transformer to provide mains isolation is therefore a most suitable compromise for the power supply module , as testing the isolation of a custom produced transformer for a High frequency ( 25 - 100KHz ) switching power supply is beyond the scope of this project though this would further increase the overall efficiency of the completed unit . For a 45 Volt output power supply the Input voltage from the rectifier must not drop below 47 volts ( allowing for ripple ) with the mains supply at its lowest limit this specifies the transformer secondary voltage Vac=VDC / 1.414=47 x 0.707=33 Volts , so assuming the regulation limits to the supply a 35 volt AC transformer of 500VA capacity will be suitable . As a side issue the regulator becomes more efficient with increased mains supply voltages as opposed to less efficient in the case of the normal linear series regulator . The maximum DC voltage applied to the regulator's input will be 35 x 1.06 x 1.414=52 volts the minimum being 47 volts for the required 45 volts output. In order to guarantee regulation , I later set the supply voltage down to 42 Volts after testing the power supply on FULL load in an Environmental chamber at my workplace .
Some form of overvoltage protection for the RF power devices is mandatory given the price of them £ 150 + vat per device . A suitable " Crowbar " protection circuit has been incorporated in to the main power supply to remove the supply in the event of a short circuit failure of the series pass element , this protection circuit would be common to a " linear regulator " as well . From these basic design criteria the power supply was developed using an integrated power control circuit to minimise parts count , so keeping the power supply module as compact as possible whilst still maintaining a relatively simple layout for the unit.
The whole power supply unit is cased up using sheet metal to screen it from any unwanted signals ( RF ) from the amplifier as well as to prevent any undue signals being radiated & picked up by the optional receive pre amplifier . The power supply for the control board has its own separate regulator powered from the unregulated DC of the main supply . The design of the auxillary supply buck regulator was based around design details in an application note from Motorola " Designing Switch Mode Power Supplies " this gives detailed design stages to enable the correct choice of regulator along with the information to calculate component values. The main power supply uses the L4970 integrated power control circuit manufactured by " SGS - Thompson " , the rectified supply is passed via a fuse to the controller , this is operated as a " Buck " type converter to reduce the supply voltage down to the desired level ( 45 volts ) a feedback loop around the converter keeps this constant irrespective of load . ( Foldback current limiting occurs @10A) From the manufacturer's datasheet it was calculated that the filter inductor had to be around 100uH & capable of 10 A peak current this could have been hand wound , but I chose to use a commercial part from the " Coilcraft " range of power inductors as this would be more compact than using a toroidal core to wind the inductor on . The choice of smoothing capacitors for the output of the regulator requires the use of so called " LOW ESR " one's , as these are characterised to have a low reactance at the high frequencies involved in " Switch Mode " power supplies . In line with the Motorola application note ,several capacitors are used in parallel to make the required value this is done in order to minimise the internal losses within the structure of the components ( I2R ) in addition to the electrolytics used a ceramic capacitor was used to remove any residual " HF " noise components which are not removed from the supply due to the internal construction of the main output smoothing capacitors , this capacitor also removes any conducted RF signals off the supply rail emanating from the RF power amplifier. These signals when superimposed on the DC level could cause loss of regulation along with subsequent triggering of the " Overvoltage " protection circuitry . The " Overvoltage " protection circuit uses the " MC3423 " crowbar driver from Motorola to drive a 25A thyristor type BT152 to short out the Input to the power regulator blowing the fuse in the supply line to it . Throughout the design stage of the power supply use was made of the " Online " design software from SGS Thompson's website for the L4970 & associated devices to check the calculated values . The whole power supply regulator being designed to fit on a printed circuit board with only the low frequency rectifier being chassis mounted ( Heatsinking of the device). In order to minimise losses within the regulator it was decided to use 1.6 mm FR4 2oz / sq" material this means that the thermal considerations of using a PCB @ 10A can be met with fairly narrow tracks , as the cross - sectional area is still met , The use of a double sided layout with extensive earth planes for stability is also necessary to minimise the radiation of switching spikes .
In order to comply with the relevant EMC regulations concerning leakage into & from the mains supply wiring , a supply filter is necessary .This may also be incorporated as a single unit containing the mains supply protection fuse and a suitable connector ( IEC style ) for a detachable power cord . In order to keep the physical size of the power transformer to a minimum it was decided to use a " Toroidal " style of mains transformer. This type also has improved regulation of output voltage compared to the traditional " E & I " so reducing the change in output voltage between no load & full load conditions this means that the DC input to the regulator will not change too far from the nominal design values . The whole regulated power supply with the exception of the bridge rectifier & main smoothing capacitor is enclosed in a screened metal box with feedthrough capacitors to minimise any residual EMC problems from using a switching type regulator also this minimises any RF signals being picked up in the regulator's control path ( output sample network )
Circuit Diagram of the regulator
Picture of the completed Regulator module
After two years use I obtained some commercial switch mode power supplies from some " Commercial " radio transmitters and one of these replaced the original transformer & switching regulator design .With one of these configured to give 45 Volts at 10A the amplifier output rose to over 250 Watts and was considerably lighter to pick up .
This page last updated 10th July 2008