Amplifier Voodoo

I have built a few amp stages, most of them either worked or worked and were total crap. This is the first time I have designed and built a multi stage amp and have things actually work how they are meant to. I wanted this amp to be very linear and to keep harmonics to an absolute minimum. And it is clean, each stage plays nice with the next stage, there is no loading up one stage, no distortion. Obviously this is not a huge power out monster, but there is 40db of gain in there so far with more to come from a fet final stage. So far I am pretty happy with how things are going.



The complete mess 🙂

About 4 watts out.

Harmonics are -45db down or better.

Schematic for amplifier chain.



So after a little bit of tinkering, its now making 5w out. After 10mins keydown it hardly gets warm, only 6 deg C above ambient measured on the case of the FET.


Op Amp Gain Bandwidth NanoVNA

I have been thinking for a while about using high speed op amps as IF amps in a receiver. As mental as that might sound to some, it actually makes practical sense in someways. Gain is easy to set, impedance is easy to set, being that the IF is at a fixed frequency you are not worried about being broad banded and can tailor the circuit to suit by using a suitable op amp with sufficient bandwidth to do the job.

And that is where the problem lies, op amp gain bandwidth is given at unity gain, IE a gain of 1 and as soon as you start adding gain, you start losing bandwidth. This means that you need a unity bandwidth of Gain x MHz to be somewhere close and then you also need a op amp with a fast enough slew rate to deliver the waveform amplitude you desire.

Now there are what are called current controlled op amps that give much better gain bandwidths above unity, but they are kind of expensive and so that leaves using voltage controlled op amps and working around all its limitations, but as you start to get up there in unity gains above 300 MHz even they start becoming non cheap items also.

So a few weeks ago I was on one of the Chinese parts sellers just looking at all the different crap they have and for some reason I ended up in the op amp section and found an op amp with a few hundred MHz unity gain bandwidth for pretty cheap. And by cheap, i am talking in the 40c each kind of space. So i bought a few to try out.

So i built up the non inverting circuit as shown above. Which is quite simple to set the gain and the impedance’s just by changing the value of a few resistors.

This is the circuit built on the test board. While the op amp is an SMD part, its SOIC 8 so its big enough that even a dummy with coke bottle glasses could hand solder, but I am kind of slack in that regard so I used paste and hot air, i mean why not. LOL

You can see from the S21 gain plot that there is usable gain from 40m to 10m. I am not sure what that notch is, but i suspect that its an artifact from the nanovna, because a manual sweep of that section of spectrum using a function generator and oscilloscope did not show that dip.

Oh I should say that I have the gain set to 6x for this test. And slew rate was not an issue for 7MHz to 30MHz, with the op amp able to deliver 1.3v peak to peak quite happily. Below that, particularly around 80m, the op amp could not deliver much more than 500mV peak to peak. So for a 9MHz or 12MHz IF amplifier, the op amp might be a credible option.


Measuring Gain Bandwidth With NanoVNA

The NanoVNA is more than just a fancy SWR meter for checking your antenna. Its much more and a very useful tool for the home brewer building amps and filters and the like. Now i have been buying things like crazy for various projects that I would like to build in the future. Often these parts are spec’ed for bands not of interest to me. So what do you do? Well you measure them at the frequency of interest and see how they work yourself.

Below is a MMIC amplifier part that I found for really cheap. By cheap I am talking in the 40 cents per range, so i bought 100 of them. I mean why not, they are spec’ed for 100mhz to 3gig with +30db of gain. Worst case senario they are kind of useless at HF and I will have parts for when I actually want to build things at VHF and up.

So anyway I had a test board built with 4 different circuits on it for testing out various parts I have here and it includes Op Amp, Mosfet and BJT amp circuits. So i decided to start with the MMIC and see what it can do.

The good thing about MMIC gain blocks is the fact that they have such a low parts count. 2 blocking caps an inductor, bypass cap and gain setting resistor. Initially i set the bias resistor a little to high and was getting a lot of distortion, so I halved the value and boom it was providing 24dB of gain at 7MHz, which is my go to frequency for all these sorts of tests. Next though, I wanted to see how the gain bandwidth was. My bandwidth of interest is HF so 3 to 30MHz, so just for shits and giggles I measures it out to the 6m band.

The test setup was Port 1 of the NanoVNA to the input of the test board, the output of the test board to the RF Sampler i made the other day, which was also connected to a dummyload and then back to Port 2 of the VNA. Then an sweep of was performed and the S21 Gain was measured.

Things actually looked quite nice and rather flat. A few dB down at 80m and 1 dB down at 6m. That is pretty good for a part with a minimum frequency of 100MHz. So all in all, this part is a winner and something I can use in a project sometime soon.


RF Sampler

Many pieces of test gear required quite low input levels, in many cases that is 0dBm (0.001W) or less. So what do you do then if you want to characterize the bandwidth of an amplifier that is putting out say a couple of volts peak to peak? You can hope like hell that you do not blow the front end out of your test gear like that expensive spectrum analyzer, you could use an attenuator, but that is going to be limiting as power levels go up beyond a few watts, or you can use a RF sampler of some kind.

Now there are some pretty fancy RF taps and samplers out there costing many hundreds or even thousands of dollars and quite frankly while not out of my budget, buying a second hand tap off an ebay seller for a few hundred bucks to use on my spectrum analyzer that cost a few hundred bucks, seems like overkill to me.

An RF sampler is not a very difficult item to construct at home, a bunch of resistors and maybe a DC blocking cap and Bob is your mothers sisters brother. Sure there are other methods and I know that the fan boys will be frothing at the mouth about using a ferrite toroid and how they are superior hams because one time they bought a resistor yada yada yada, you know the instant internet types.

Anyway, I settled on the resistor divider method with a bit of series and parallel magic and some 1 watt resistors and the magic of the resistor divider not dissipating all the power you get a pretty low cost way of sampling higher powers and the ability to set the amount of attenuation by paying with the resistor values.



Here is the circuit that i designed in my ecad to make a pcb board.

The astute will notice that I am not using the same value resistors on this board as in the schematic, turns out i used the footprint for the 1w resistors that I have and well had to use 1/2w 1500 ohms instead. The value is kind of non critical anyway it just changed the amount of attenuation.

And finally, here is a S21 gain plot for the HF ham bands. Not quite linear, but -45 to 47dB. Good enough for the kinds of girls I go out with. And its something I will be using quite a bit as I also have a test board I had made so I can test some parts out to see how they perform. First one will be an high bandwidth op amp that might be suitable to use an an IF Amp stage. More on that soon.


9mhz IF Filter

Well i did not think i would get any bench time so soon, or feel enthusiastic for that matter, but here I am tinkering today and doing McRadio at the McBench. Anyway, i think i posted about this IF filter sometime back, I should go and look but I wont LOL, anyways its a Cohn Min Los design and what I did differently was not to use transformer matching, but rather to use LC impedance matching to get the 50ohm up to the 200 ish ohms of the filter.

Now i did also build one using transformers, the difference was, this one has less of a stop band but better shape and much more symmetrical, the other one had a better stop band but its shape was not very good at all. Greater attenuation would be nice and I think the reason for the lower stop band is the Q of the LC match, which is going to be rather low, somewhere lower than 10. After doing some reading, I thing by using a T or Pi match I can improve the Q a lot and that would improve the stop band. Its an experiment i plan to do sometime soon to see if it improves things.

This is the PCB board i laid out and had made, the xtals I used were just whatever 9mhz junk xtals I could stuff in there. No matching, nothing, just grab 8 xtals and stuff them in. The block of wood is the former I used to make the shielding can.

The shielding can is made from some 0.1mm brass sheet. I placed it over the former, clamped it in place with a g clamp then beat the crap out of it with a hammer and soldered it in place.

50db stop band is not all that great, but it should be usable. Pass band is a little lumpy, but should also be acceptable. The shape is quite nice and could be used upper or lower side no problems at all. the width is 2300hz. Good enough for the kinds of girls I go out with, as they say.