E类功率放大器设计.doc

《E类功率放大器设计.doc》由会员分享，可在线阅读，更多相关《E类功率放大器设计.doc（12页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、E类功率放大器设计 Alan Melia G3NYK, Mike Probert GW4HXO, and Finbar OConnor EI0CF艾伦苦G3NYK，迈克普罗伯特GW4HXO和Finbar奥康纳EI0CF There have been many different configurations that have attempted to squeeze the best efficiency out of an RF Power Amplifier.已经有许多试图挤进了一个最佳的效率射频功率放大器进行不同的配置。 Class E is a form of switching

2、amplifier which was patened by Nathan Sokal WA1HQC in around 1976. E类是开关放大器的形式，patened由Nathan索卡尔WA1HQC周围一千九百七十六英寸 I first saw it described in Design Electronics in 1977.我第一次看到它在设计电子所述于1977年。 Being an amateur, Nat is an extremely practical man and the article gave a test circuit that could easily be

3、assembled from laboratory components, and then measured whilst the components were altered.作为一个业余爱好者，NAT是一个非常实际的人的文章介绍了测试电路，可以很容易地从实验室组装部件，然后测量虽然成分改变。 I had little interest at that time but the article was filed away until I had attempted to modify an audio amp for 73kHz with very little success.我当时

4、的兴趣不大，但被束之高阁的文章，直到我曾试图修改一个非常小的成功为73kHz音频放大器。 I dug out the old article and read it several times.我挖了旧的文章，读了好几遍。 It did not give any guidance about where to start from in a new design, though it gave a number of optimised equations for calculating the critical component values.它没有提供任何有关从哪里开始从一个新的设计，虽

5、然它给了关键元器件的计算值的优化方程的数目指导。 I started by putting the equations into a spreadsheet and tried some numbers to see what changed, with power, supply voltage and load.我开始通过把电子表格的公式，并试图用一些数字来看看有什么改变了电源，供电电压和负载。 As a result I think I gained a feeling for the circuit and developed a recipe for the process of p

6、roceeding through a design因此我认为我获得了该电路的感觉，制定了通过诉讼程序设计一个配方 The following notes are a result of experimental work, and in no way add to the theory which has been adequately covered by Nat and Fred Raab in many papers.下面的说明是一个试点工作的结果，绝不添加到已获得足够的NAT和弗雷德拉布多篇论文涵盖理论。 My objective was to try and formulate a

7、n easy route to a medium power 136kHz power amplifier, which would be relatively benign.我的目的是试图制定一个简单的路由到中等功率136kHz功率放大器，这将是相对温和。 By benign I mean a circuit that did not have an insatiable appetite for FETs.我指的是由良性电路，没有一个场效应管贪得无厌的胃口。 The RSGB LF reflector has carried stories from time to time of exp

8、loding FETs and of complex protection circuits to circumvent this problem.低频反射的RSGB已进行的故事不时爆炸场效应管的时间和复杂的保护电路来绕过这个问题。 My naive thinking was that a switch which is either on or off should form the basis for the circuit, using a single ended design to avoid the cross-coupled oscillator effect.我天真的想法是，这

9、是一个开关打开或关闭应成为电路的基础上，采用单端设计，以避免“交叉耦合振荡器效应”。 It seemed that some LFers had already been playing on the verge of Class-E but there was no information with the circuits and it seemed they did not really know how it worked or how to set it up.这似乎有些LFers已经对E类边缘打，但没有与电路资料，而且似乎他们真的不知道它是如何工作和如何设置它。 Unlike ma

10、ny PAs that we are familiar with, you cannot just increase or decrease the power from a Class E PA by varying the loading on it.不像我们所熟悉的许多功率放大器，你不能只是增加或减少其上的负荷变化从一个E类功率放大器的力量。 The first important thing to appreciate is that a Class E PA is designed for a particular power output, and it will only be

11、efficient at this design power level.首先重要的是要明白的是，一个E类功率放大器是为特定的输出功率设计的，它只会在这个设计功率效率。 Click here to download Excel worksheet http:/www.alan.melia.btinternet.co.uk/download/classe.zip点击这里下载Excel工作表http:/www.alan.melia.btinternet.co.uk/download/classe.zip Design proceedure设计过程 1. 1。 Select a Power Supp

12、ly voltage (Vdd).选择一个电源电压（Vdd）。 2. 2。 This determines the Vdd(max) of the FET at approximately 3.5 times Vdd这就决定的VDD（最大值）的约3.5倍，在Vdd的场效应管 3. 3。 Select the target power level选择目标功率水平 - This determines the optimum value of C1 and the optimum load impedance -这就决定了C1的最佳值，最佳负载阻抗 4. 4。 Determine the the v

13、alue of C2 and L2 from the Excel worksheet确定从Excel工作表中的C2和L2的价值 Nats original paper didnt give guidance but his later QEX paper suggests that the value of the choke L1 is not critical, but it should have an inductance with a reactance at least 30 times the optimum load impedance. Nat的原始文件没有给予指导，但他后来

14、QEX文件建议，该扼流圈L1的并不重要，但它应该有一个电抗至少30倍的最佳负载阻抗电感。 this means at 136kHz a minimum of around 500uH.这意味着在136kHz左右的500uH最低。 I have been using a choke of 5mH as I want to operate at 73kHz as well.我一直在使用的5mH呛，因为我想在73kHz运作良好。 Also I suspect that the larger choke will give a softer keying risetime.此外，我怀疑大呛会给一个软键

15、控上升时间。 Now for some practical details.现在对一些具体细节。 The point of Class E is that there is no power wasted other than the Idss*rds(on).在E类的一点是，有浪费比其他的RDS的IDSS *（上）没有权力。 This is achieve by the series circuit C2L2.这是实现由串联电路C2L2。 Because C2 will have to be made up from standard values of pulse rated HV cap

16、acitors it is necessary to have some adjustment on L2 for initial set up.由于C2的将须从高压脉冲电容器额定标准值的注册必须有一些初始设置二级调整。 You will see in the worksheet a column which calculates the resonant frequency of L2C2, and may be somewhat surprised to see that it differs from the driving frequency.你会看到一列在工作表中，计算了L2C2谐振

17、频率，并可能有点惊讶地看到，它从驱动频率不同。 This is because this circuit is NOT a tuned circuit to extract the power at 136kHz.这是因为这个电路是不是一个调谐电路在136kHz提取的权力。 It is part of a flywheel circuit to ensure that the correct voltage waveform can be generated at the FET drain, during the FET off period.这是一个飞轮的电路，以确保正确的电压波形产生于在

18、场效应管FET的漏极关闭期间的一部分。 The drive waveform for maximum efficiency should be a 50:50 squarewave of about 10volts amplitude (see the drive circuits on Dave G3YXMs PA and Dave G0MRFs design).为最大效率驱动波形应该是一个幅度约10volts 50:50方波（见戴夫G3YXM的PA和戴夫G0MRFs设计的驱动电路）。 Dropping close to ground on the off periods.删除接近地面上的关

19、闭时间。 The circuit is matched to 50 ohms by a ferrite transformer.该电路匹配50欧姆的铁氧体变压器。 the rule-of-thumb for ferrite transformers is to make the (unloaded) reactance of the winding at least three times the impedance it is fed from.的原则进行的铁氧体变压器拇指是使绕组的至少三次的阻抗这是美联储从（空载），电抗。 Thus if the optimum load impedanc

20、e R is given as 10 ohms by the worksheet, the inductance of the primary should be 10*3/2*pi*f.因此，如果最佳负载阻抗R是由为10欧姆表给出了初级电感应该是10 * 3 / 2 *圆周率*楼 Because we are only interested in a narrow range of frequencies, I have used 5 times rather than 3 times.因为我们只是在一个狭窄的频率范围有兴趣，我已经使用5次而不是3倍。 This makes the seco

21、ndary taps a little finer, and makes it easier to find the maximum power position.这使得第二水龙头有点细，并使其更容易找到的最大功率的立场。 The vital part of the set-up proceedure is to monitor the drain waveform on an oscilloscope.该设置程序的重要组成部分，是监察流失在示波器上的波形。 As the FET switches off the current is maintaned by the choke and st

22、arts to flow into the capacitor C1.由于FET开关关闭电流maintaned的呛，并开始流入电容C1。 As the capacitor C1 charges up current starts to flow through the load via the series circuit C2L2.由于电容C1开始收费，以电流流过负载通过串联电路C2L2。 After the voltage across C1 reaches about 3 times the supply voltage the load draws curren from C1 and

23、 the voltage starts to fall.之后，在C1上的电压达到约3倍的电源电压，负载消耗柯伦从C1的电压开始下降。 To achieve the maximum efficiency the voltage across C1 should fall to zero by the time the FET is switched on again.为了达到最高效率应该在C1上的电压下降到零的场效应管再次开启的时间。 This ensures that the FET does not wastefully discharge C1.Any discharge by the F

24、ET at this point of the cycle does not generate any RF power in the load.这确保了场效应管没有在这个周期点浪费放电C1.Any由场效应管放电不会产生任何负载的RF功率。 It is even more complex than that for the slope of the voltage waveform should also flatten to reach zero volts, with a kind of soft landing.它更比电压的波形斜率，复杂的也应扁平化，达到了一种软着陆零伏。 This i

25、s the condition that will yield maximum efficiency.这样的条件下，将产生最高的效率。 This waveform and how it depends on C1 and C2 and L2 is described in the QEX article Jan.Feb 2001 pg 9.这个波形，以及它如何在C1和C2和L2中描述取决于QEX文章Jan.Feb 2001页9。 Class-E RF Power Amplifiers by Nathan Sokal. “E类射频功率放大器”由Nathan索卡尔。 I have achieved

26、 175 watts for a DC input of 190 watts (38 volts at 5 amps) using a single IRF640 which is a cheap TO220 packaged device.俺都已经达到了190瓦直流输入使用一个单一的IRF640这是一种廉价的TO220封装器件（38个5安培伏特）175瓦。 The drive was not keyed for long but the FET did not seem to show any signs of distress, despite a very small heat sink

27、.驱动器尚未键长，但似乎没有场效应管遇险任何迹象表明，尽管一个非常小的散热片。 The IRF640 is a 200volt device so the absolute supply limit is about 55volts.该IRF640是200volt设备供应的限制，因此绝对是关于55volts。 Mike GW4HXO has also achieved similar levels and is constructing a dual (parallel) FET 300 watt version on a sizable heat sink.迈克GW4HXO也达到类似的水平，

28、是构建双（平行）场效应管300一个相当大的散热片瓦版本。 Finbar EI0CF has used a different configuration but has achieved powers of about 530 watts, at efficiencies that we think, allowing for RF ammeter inacuracies must be over 90%. Finbar EI0CF公司采用了不同的配置，但已取得的效率，我们认为在约530瓦，权力，允许射频电表inacuracies必须超过90。 Finbar uses two 500volt

29、IRFP devices (better heat sinking.more expensive) and a supply voltage of up to 90 volts. Finbar使用两个500volt IRFP设备（更好的散热.更贵）和高达90伏的电源电压。 He also used the scope trace of the FET drain waveform to optimise the set-up, despite a totally different reactive network.他还使用范围的痕迹FET漏极波形优化设置，尽管一个完全不同的反应网络。 Mik

30、es and my experiments have been carried out into dummy loads only but Finbar has fired the 500 watts into an aerial (I dont think he has a dummy load that big .yet.)迈克和我的实验已经进行，但Finbar已成为唯一的空中发射了500瓦的假负载了（我不认为他有一个假负载，大.这个问题。） The one pleasing thing is that none of us has lost a single FET in these e

31、xperiments.一个可喜的事情是，我们都已经失去了在这些实验中单FET。 The worst condition is probably a short circuit load, any other condition seems to reduce the power drawn by the FET.最糟糕的情况可能是负载短路，任何其他条件似乎可以降低由场效应管得出的权力。 Even a very low load impedance seems to damp the circuit somewhat.即使是非常低负载阻抗电路似乎有些潮湿。 It may seem dauntin

32、g to have a scope attached to the TX all the time.它似乎令人生畏有附加到TX的所有时间范围。 This is really only necessary in the early stages of set up.这实在是只需要在成立的初期阶段。 Once the circuit is set with the correct component values.一旦电路设置正确的元件值。 the PA can just be switched on and keyed.巴勒斯坦权力机构可以直接接通和键控。 If it is necessary

33、to tune the aerial.如果有必要来调整天线。 The supply voltage can be reduced, this does give sub-optimum efficiency but does provide reduced dissipation as well.电源电压可以降低，这确实让分，但并提供最佳效率，以及降低功耗。 The PA components C1 C2 and L2, and the transformer tap should NOT be altered but the loading coil should be tuned for

34、maximum aerial current. * addition 7/2/02* Johan SM6LKM has suggested that it is not quite right to tune for maximum current as the maximum output power of a Class-E stage does not coincide with the condition of maximum efficiency.该PA元件C1的C2和L2和变压器抽头不应改变，但目前的装载线圈应调整最大天线。* *此外7/2/02约翰SM6LKM建议，这是不太正确的

35、调整最大的一类- E的阶段，最大输出功率不配合的最大电流效率状态。 The alternative would be to use a bridge such as that described by Jim M0BMU in the LF Experimenters Handbook to adjust the aerial to a true resistive match.另一种方法是使用一桥如由吉姆在低频实验者手册M0BMU所述调整天线到真正的电阻匹配。 Another aternative is to use the simple transformer bridge to adju

36、st the aerial tuning first see my aerial measurements page *另一个aternative是使用简单的变压器的桥梁，调整天线调谐第一眼看到我的测量天线页 * If aerial matching is employed this should be adjusted to present the transformer secondary with 50 ohms.如果采用这种天线匹配应调整目前的变压器50欧姆次要的。 The supply may then be switched up to full voltage and the P

37、A keyed.随后的供应可能被切换到全电压和巴勒斯坦权力机构键入。 If Q values of around 10 are used for the flywheel circuit (see the worksheet) the PA will not need any adustment across the whole of the 136khz band.如果Q约10人的飞轮的电路中使用的值（见工作表）巴勒斯坦权力机构将不再需要横跨整个adustment 136khz乐队。 It may be possible to switch C1 and C2 and use the sam

38、e coil L2 (differnet tap) to use the PA on 73 kHz as well (at least that is what I am hoping)它可能会转C1和C2和使用相同的线圈L2（目前存在的自来水）使用73千赫以及巴勒斯坦权力机构（至少这就是我希望） There is a lot written about Class-E and its variants.有一则关于E类及其变种书面很多。 It is used right up into the microwave bands for FM and digital TX stages.这是正确的

39、使用成调频和数字微波频段发送阶段。 I have just 6 papers, and they are covered in mathematical squiggles.我刚才6篇论文，它们是在数学花体覆盖。 Fortunately we dont need to be able to understand the Maths to use the idea !幸运的是我们并不需要能够理解使用数学概念！ I have also recently used a simulation program called SIMetrix (a free download from NewberryT

40、ech) to plot waveforms for different circuit values, and have found this very instructive.我最近还采用了所谓的SIMetrix仿真程序（从NewberryTech免费下载）来绘制不同的电路值波形，并发现了这个非常有启发性。 TAILPIECE附属物 One would expect a square wave drive PA to be horribly dirty with harmonics, the Class-E stage is NOT.人们会想到一个方波驱动器与功率放大器是可怕的谐波脏，E类

41、阶段则不是。 If we use a design Q of 8 to 10 the output waveform only has about 1% harmonic distortion, when set up properly.如果我们用一个设计Q的8至10输出波形只有约1的谐波失真，当设置正确。 I think this is due to the optimum adjustment of the flywheel circuit.我认为这是由于飞轮的优化调整电路。 I believe with a high Q LF aerial, this almost makes furt

42、her filters unnecessary.我和一个高Q低频天线相信，这几乎使进一步过滤不必要的。 I need to do more work with the analyser before I will stand by that assertion, though.我需要做的分析之前，我会站在这一断言更多的工作，虽然。 The graphs below were produced from the simulator SIMetrix.下面的图表是产生于模拟器的SIMetrix。 The first shows the drain voltage and drain current

43、 with too little inductance in L2.第一个显示的漏电压和漏太少的L2电感电流。 Note the big spike of current as the FET discarges C!注意：当前大穗的场效应管discarges C处 which has not reached zero volts at FET switch-on尚未达到零伏，场效应管接通 The second graph shows the results of too little inductance in L2.第二个图形显示在L2电感太少的结果。 The load current i

44、s higher but there is a wide high wasteful spike of current through the FET at switch-on.负载电流较高，但存在着在一个宽高的场效应管的电流尖峰浪费在接通。 This current spike is just heating the FET.这仅仅是加热电流尖峰的场效应管。 Notice also the voltage waveform did not reach zero and is rising again by the time the FET switches on.还要注意电压波形没有达到零增

45、长，是由当时的FET开关再次打开。 There is about 80 volts across C1 which the FET discharges.目前在C1的约80伏特的FET的排放。 The final graph show the situation at near optimum.最后的图表显示，在接近最优的情况。 There is still a little spike at swtch-on, but the voltage waveform is down to zero and is almost flattened out.还有一个小尖峰在动开关上，但电压波形下降到零

46、，几乎夷为平地了。 Although Nat suggests monitoring FET current in his 1977 article, it would seem to be only necessary to monitor the drain voltage.虽然NAT监测表明在他1977年第场效应管的电流，它似乎是只需要监测的漏极电压。 The adjustment is quite critical, but you know when it is right.调整是相当关键的，但是你知道什么时候是正确的。 The plot below is a zoomed vers

47、ion of the waveform at the drain (green) showing that the profile is flattening out, but the gradient (slope) of the waveform is not zero when the fet switches on.该地块下面是一个在漏极（绿色），显示配置文件是平坦的，但梯度（斜率）波形的波形放大版本不为零时，FET开关上。 That is the reason for the small current transient *blue) and some of the loss.这是

48、为小电流瞬时*蓝色的原因）和一些损失。 This plot was produced later than the above and may not be for exactly the same component values, hence different current and voltage scales.此图制作晚于上述，不得为完全相同的元件值，因此不同的电流和电压表。 12设 计。表流电此件元完得上作此 些和的时瞬电是. ) 。关 时本放形）（，坦文示显绿在是下该 ) ) ( 。确候什知但关相调 漏监要是，管场 在表 然 。平夷零形压但动尖一. 。的近示表后. 。排 的0约 目 开再 由，零达压注. 。应的电热仅 。在峰流的场个着但高载. 。的电 形个 通通应零未 - 处 管穗当意