EMRFD Message Archive 10931
Message Date From Subject 10931 2015-03-26 22:53:25 kerrypwr Amplifier Design It's been a bad day here in Lake Wobegon; perhaps the amplifier gurus can help.
I wanted a bench amplifier with about ten watts output; that is provided by a kit from Kees K5BCQ.
I can recommend it; it works very well.
I want to drive it from a signal generator so it requires a driver amp. Efficiency is of no consequence; I want a broadband amp that is working well within its capabilities and therefore produces a clean output without filters.
A minimum amp gain of 20dB is required.
I designed this;
http://postimg.org/image/vjpi9meet/
and it worked very well; output is very clean up to about 20dBm when it begins to compress.
Here is the completed amplifier block using the driver schematic above;
http://postimg.org/image/hsety2s8l/
The performance, as shown on the label, is for the driver alone (lower trace) and for the complete block (upper trace).I could use about 25dBm before compression so I re-designed the driver amp;
http://postimg.org/image/s17eepadt/
This does not work; it seems to have only a few dB gain.
I've checked, re-checked and checked again; I can't find a wiring error or a wrong-value component.
Have I made a simple design error somewhere?
If so, it should be immediately obvious to someone who has more knowledge & experience than I have.
Kerry VK2TIL.
10932 2015-03-26 23:36:04 Russell Shaw Re: Amplifier Design 10933 2015-03-27 00:00:03 kerrypwr Re: Amplifier Design Thanks Russell.
I thought I'd try the separate choke arrangement in the second design; no real reason, just something different.
I will either reduce the turns on the choke or revert to the "transformer - as - choke" arrangement of the first design.
I will not be able to get back to the bench for a few days so it might be next week before I can try this.
Kerry VK2TIL.10934 2015-03-27 00:02:07 Russell Shaw Re: Amplifier Design 10935 2015-03-27 00:06:27 Russell Shaw Re: Amplifier Design 10936 2015-03-27 00:13:45 kerrypwr Re: Amplifier Design Thanks Russell; our messages crossed.
I will do some breadboarding based on your most helpful suggestions.
Frequency range is determined by the amplifier that follows this one; it delivers useful power from about 3 MHz to about 50 MHz so the driver should do this at least and, preferably, a little better.
Kerry VK2TIL.10937 2015-03-27 01:02:06 Russell Shaw Re: Amplifier Design 10938 2015-03-27 02:41:45 kerrypwr Re: Amplifier Design Thanks Russell.
Your comments are well above my knowledge level; I will have to digest them very slowly and hope that I can eventually understand them.
But it's all about pushing my understanding to another level; it just takes me some time!
Regards, Kerry.10939 2015-03-27 08:16:51 kb1gmx Re: Amplifier Design My question is simpler...why 150mA? What power out is needed or desired? That's the kind ofcurrent I'd use for a few hundred milliwatts of class A output.If you doing a pre-driver for the K5BCQ amp I'd expect you want to bein the 10milliwatt range.However yes the collector choke is likely saturated but if that werethe case you would see increasing power with increasing frequency.Further at 150 mA the output impeadance is not 200 ohms likely closerto 80 ohms then you step it up by 4 so the load match is around 320ohms which when loaded ot 50 ohms will have a poor power transfer.Added to that the input impedance is likely so low at that current(under 10 ohms) that the input is badly mis-matched and that resultsin loss as well.Drop the current to 50 ma. Reduce the turns on the collector choketo maybe 10. forget the output transformer. At that current youcan drive 50 ohms direct to about 100mW.NOte. for stability and good design if you need 20DB and higherpower (more than 50mW) its better to divide the gain across two stagesof modest gain and also the first stage running at lower currents.The values chosen for the feed back resistors with a 50 ohm source willnot give 20db of gain. Two reasons, too much feedback, too low a base(input) resistance at high currents. Off the top of my head I'd expectmaybe 10 db of gain but it could be as low as 6-7.Allison10940 2015-03-27 08:47:26 kb1gmx Re: Amplifier Design Russell nailed it.To get that high a gain at 50mhz you need to keep the collector and base impedancelow but that makes wide band harder as you need transformers for matching. Ferriteslike FT43 are not all that great at 50mhz and FT61 is not so great at 1.8mhz so caremust be used and applied knowing this.Keeping with the above you also have capacitances at 50mhz that while small invalue will be significant as yo go up in frequency. These force lower impedanceas that are now R and X. Another way to say this is 4pf at 50mhz is an Xc of795ohms in parallel with the load resistance, at 10 ohms 4pf is close toinsignificant at 50mhz at 200 ohms its there and not even 4x the load.To get 25dbm the load impedance is in the 35 ohm range maximum. If the emitter has 3ohms you can expect a gain of around 10 DB. As the 3ohm resistor is part of the feedback network being unbypassed. Most amp designers at that power leveland bandwidth that use resistance in the emitter watch the amount used as toohigh will kill you and also hurts the total available voltage swing and can cause early saturation.Also at 50mhz I do not go for 20DB of gain in one lump as doing that at power (nearly ahalf watt) is beyond bipolars at 12V can do and more in the ballpark of 48V RF LDMOSfets. At 50mhz I"d shoot for more like 11dB of gain and prefix that with a lower currentstage that can do 10dBm at about 10 to 13dB. Excess gain is easily fixed with a attenuator.That type of amp does not scale to high power that being more than about .5 to 1Wwithout care in currents and resistive losses without resorting to 24 volts or higherand also accepting the gain will not be near 20dB. In the end we want emitter resistance for DC stability but work hard to bypass it for RF as any resistance there sets maximumgain.Allison10941 2015-03-27 12:51:47 kerrypwr Re: Amplifier Design Thanks so much to both of you for the "tutorial"; this is the kind of clear information that I can't find in books or on the 'net.
Even EMRFD and SSD don't go into this level of description of the underlying concepts so I've been having a great deal of trouble trying to understand amplifiers.
Kerry VK2TIL.10942 2015-03-28 08:42:59 kb1gmx Re: Amplifier Design HI EMRFD does address this.. Just not i one place as its a combinationsof multiple interacting topics.Transistors and their models.Amplifier gain, specifically feedback amplifiers.Impedance matching Why and How.Several areas where power amplifier design is addressed.Your getting caught up in impedance matching and power amp design.There are no simple axioms for that. As power goes up device interactionsalso manifest themselves in greater ways and impedance matchingbecomes more critical. As power goes up impedance goes downat the device level. Also as frequency goes up impedance goeslower.If we add circuit capacitances and inductances both of which arealways present we get a more complex picture. FYI at 3mhz thoseare less noticeable than at 50mhz. For example a .0032uh (3.2nh)(about 1.4" of wire) at 50mhz is a 1 ohm reactance, at 3mhz that wirewould have be maybe greater than 10 inches long. [Making a pointof lead lengths and its effects.] So leads at VHF even a tiny .25" onescount. And at high power where impedance are low adding a half ohmof reactance is a big deal. Which is why even low power (1W) devicesat VHF and UHF have wide very short leads. Its also why at VHF SMTas taken circuit design by storm, less leads, less parasitic inductance,and capacitance. Also if the circuit impedance is around 10 ohmsyou can see that 1 ohm of reactance added to that can have significanteffects at VHF and not even seen at low HF.How does one learn about this... Look at other works from others[both commercial and hobby] that do work and study them. Understandwhy they ended up with what they got. this is both circuit, layout, andconstruction. Those three are often only have a remote relationship.The schematic is where the wire ends meet but nowhere does atypical schematic state this wire is .3" long. [exception is in VHFand power circuits drawn for reproducability.). Then we have alayout drawing, usually how a PCB is to be made. Then there isas built and made working where areas are jumperd with coppersheet and through wire added to correct errant RF current flows.The study part is why did they use that cap value or that resistorvalue and so on. Same for transistors used. Sometimes its justthat part is in stock, handy, or preferred. Other times its a specificpart with specific capability.I'd suggest going back and looking at what was done for a driverstages in EMRFD and other places (the 'net!) and try and understandwhat others did and compare to what you built.This might help... the IRF510 and the RD16HHF both net about 13DBof power gain. so any driver for either will need to produce about .5W[maybe more to allow for losses] for 10W output.Its not as bad as it sounds but you do have to look.Allison10946 2015-03-29 02:13:25 Russell Shaw Re: Amplifier Design 10949 2015-03-29 19:56:35 kerrypwr Re: Amplifier Design Allison is (almost) correct; EMRFD does address many (but not all) of the details but in various locations.
I went through the exercise recently when trying to improve my very basic understanding of amplifier design.
I scoured EMRFD, SSD and even, braving the equations IRFD; each, supplemented by extensive internet trawling, gave a piece or pieces towards completing the puzzle
I wish that I could find a comprehensive article on the subject.
It seems to me that the most difficult design task relates to amplifiers of about one or two watts; that is where all the interdependent variables become very sensitive.
Lower-power amplifiers aren't too difficult; I have managed to design and build those with no drama, at least after the first few. :)
Although I have no experience with higher powers, they don't seem too difficult either; perhaps this is in the realm of "famous last words" :) but I think that the separate bias supply eases a lot of the difficulty.
But it's in the middle ground, that one or two watt area, where separate bias supplies aren't worthwhile and tiny variations in a component value can upset one or more parameters, that life becomes difficult.
So far as the K5BCQ amplifier is concerned, it has about 20dB gain over HF; to produce 10 watts/40dBm it requires about 20dBm drive. Twenty watts requires 23dBm drive.
My original design gave 20dBm (with about 0 dBm input) with minimal harmonics but quickly got worse after that point; as my original desire was for a "clean" 10 watts without filters I will re-install that as I want to use the amplifier.
I wanted this output in the HF range but the K5BCQ continues to produce useful gain up to 50 MHz so I thought that the driver should complement that very good performance.
But I will work on a replacement; two stages, as Allison suggested, would be preferable.
The first stage should not be too difficult for me; the second stage will, I suspect, teach me a good deal!
Thanks to you both for the "chat" and the very constructive help.
Kerry VK2TIL.10955 2015-03-31 20:37:04 kb1gmx Re: Amplifier Design I still maintain that EMRFD has the needed info. I didn't say it was easy to findor all in one place.Amplifiers be they low power receiving or dummy load heaters have all the same issues.Some are less profound or more so depending on power and voltages.>>I wish that I could find a comprehensive article on the subject.
They do not exist. The topic cover the full range of engineering.Yes. At low power you can use feedback to force the stage behavior. As you go up in
>>It seems to me that the most difficult design task relates to amplifiers of about one or two watts; that is where all the interdependent variables become very sensitive.
power then increasing emitter current leads to lower device input resistances.Its the range where I likely spend more time as well developing and testing.>>Lower-power amplifiers aren't too difficult; I have managed to design and build those with no drama, at least after the first few. :)
Generally up to 10 to 20dbm they can be done by formula. Above that they do tend to depart from predicted unless one goes empirical with testing as small signal formulaare not sufficient.You'll eat those words.. ;) 45 years of dealing with it says it will get you if you don't pay serious attention. I know a lot of people doing 5W and up with IRF510s that found that
>>Although I have no experience with higher powers, they don't seem too difficult either; perhaps this is in the realm of "famous last words" :) but I think that the separate bias supply eases a lot of the difficulty.
out. Try 8 of them push pull at 50mhz and 250W... yes they do!
The separate bias thing is because self bias is impossible. consider a 100W amp at 12Vwith a aggregate collector current of 20A. Any resistance in the collector or emitter isgoing to eat all your power never mind gain. Also at those currents you end up withinput impedance in the 1-2ohm region, the outputs are also similarly low or lower.
>>But it's in the middle ground, that one or two watt area, where separate bias supplies aren't worthwhile and tiny variations in a component value can upset one or more parameters, that life becomes difficult.
First is you can external bias. Also as you get into the 1W range adding emitter resistance is desirable for DC stability but even a ohm will impact gain and output power significantly.That is why your first amp was good but crashed had when you went for more power out.The transistor was literally running out of voltage and current. At the 1-2W out range things like emitter resistors and resistors in the power source line are significant to voltage loss.They need to be as small as possible. For the DC source a choke might do the task for decoupling the stage. For the emitter maybe an ohm (1) and if you want more than 10DB gain none. Feedback is likely hard as the values due to the input resistance will beabsorbing power (more gain reduction than calculated). Then you have thesmall signal input impedance equation Rin=26*B/ma where B is the ac beta at the chosen frequency and ma is emitter current. Right around 1W its not quite accurate as internal resistances are taking significant form ( we will ignore reactive component for now butit also works against you) but at 100mA and 20mhz for a high FT device that works out to maybe 5.2 (assume beta of 20 at RF) and that's ohms. The implications is thatwith that current and a 2 ohm emitter resistor (unbypassed) the gain will be low as the current is across 7 ohms only 5 of which are the BE junction resulting in loss of gain.at 1 ohm and 1W that may be tolerable if you can live with maybe 11DB of gain withoutshunt feedback, DC stability may require it. If its bypassed (hard to do at 2 ohm) itsstill a DC loss of .2V of DC.Every part of the amp those items get more significant compared to the device. If the load resistance is to high you get limiting, too low low gain and less power than expected due to load mis-match. If the power isn't transfered do to that mis-match then heating is an issue(and possible damaging over current or voltage swings).All of the elements are interactive and combinational. Some conspire to make somethings hard or very difficult to do like high gain.Allison/KB1GMX10956 2015-04-01 22:07:11 iq_rx Re: Amplifier Design Here's a shameless plug. I've been living with EMRFD for longer than anyone except Wes, and had read it cover to cover several times before it went to print. I am amazed that each time I re-read a section, I uncover depth that I missed in the first decade or so. EMRFD was Wes's third book, and by then he was uniquely skilled at covering the entire space from introductory circuits to technical problems that baffle the most experienced professional designer--often in the same schematic! A novice can get it working successfully, and a professional can dig deeply into the subtle interactions between components and still generate more questions than answers.Regarding PA design...Allison nailed it. Every frequency range and power level has its own opportunities and constraints, and moving from excellence in one to the next is perhaps a bit like designing a formula 1 race car, starting with a Tour de France bicycle. In the words of the great designer Phil Bolger:"There's a lot to be learned from studying this design, but to apply the lessons you have to start over with a blank sheet."Back to EMRFD, it's worth a 2nd, 3rd...nth read, and often an answer to a puzzling question may be found while studying a design from an earlier or later chapter.Best Regards,Rick KK7B10957 2015-04-02 12:38:33 p1jn3v11 Re: Amplifier Design I have for sales 3 sets of W7ZOI Cascode IF Kits please see my ad in eham classified under QRP
Thanks
Steve
KF7ZFC
10958 2015-04-02 18:59:52 David J Nushardt Re: Amplifier Design Hi Steve , I looked but could not find it, I may be interested in purchasing a kit, could you email me the link, thanks much!
73's Dave
N9OOQ