Wideband Amp

[T3sl4co1l]

I don't want to derail the other thread, but it sounds worthy of discussion, so here:

Are you sure you get a flat bandwidth up to 700MHz out of that module?

This is what it looks like after debugging:
https://www.seventransistorlabs.com/Images/WidebandAmpDebugged.jpg
(bottom left of the scratch paper is relevant, for once)
Full schematic:
https://www.seventransistorlabs.com/Images/WidebandAmpSchem.png

Inductors with "xxxR" are resistors of that value, and likewise for resistors in nH.  Components shown with a resistance have that added in series (except for L6, that's just the DCR of the real component, a Coilcraft 1812CS part).

This version has lots of ESR added to dampen things.  I made another more radical variant that's studded with more bypass caps, so I'd mixed up my descriptions there a bit.

Frequency response was measured with a not-very-flat noise source, and measuring the difference of averaged spectra.  While the spectrum isn't very flat, I have no particular reason to doubt the gain is more or less correct: it's not in compression, or below the noise floor.

It scares the bejeezus out of me. Have you ever looked into it with a VNA? I know you have cured the instability but I'd expect to see spikes of negative resistance up above 2GHz from what I see in the version on screen.

Could be.  I only have the spec (1.5GHz, but it actually stops at 1.8 or 2), and no signal generator or return loss bridge to work in that range.

Even if you grounded the base really well the BFR92 itself has parasitics and the output signal path looks quite long before it reaches the next BJT stage. There isn't much ESR here and I'm not sure what the input Z of the next stage is at 2GHz but there's enough trace length to do a decent trip around a smith chart at UHF so I'd expect to see negative resistance at the board input 'somewhere' in the upper UHF region and maybe beyond this. But I can only look and guess. I think other engineers at my place of work would also be scared of that module in a review

Good thing it's just for sitting on my bench and (hopefully) amplifying signals in the given range without trouble.  Well, apparently not much trouble, in the range that I can see...

But even in those days I don't think a wideband (UHF) MMIC alternative would have been tolerated in a design review because of the risk of instability up at many, many GHz. i.e. at frequencies beyond the manufacturer's s parameter tables. There's nothing like putting a lid over a marginal amplifier design and then cooling it to -40degC to make it wake it up into oscillation

The other amp to which I referred (just now in the thread) is this,
https://www.seventransistorlabs.com/Images/20MHz_LNA.jpg
https://www.seventransistorlabs.com/Images/20MHz_LNA_Simpl.png (simplified, missing some impedance and frequency equalizing components)
It's around 3.2nV/rtHz in a 20MHz BW, which is obviously not very low at all, and nowhere near as low as the puny BFR92 is doing.  It could be PBSS303NX is just not low noise (but it's a comparable device to the ZTX751, also unspecified for noise but happens to be good), but it seems more likely I've forgotten something gross about the circuit.

Red herring: the mirror bias network isn't it.  I can replace it with a 150R resistor (base to GND) and get the same noise...

Tim

[G0HZU]

OK thanks for the extra info. It still looks a bit scary with the 2p2 cap in the cascode emitter. I'd expect to see a big and sharp gain spike up in the UHF region where the microstrip and the 2p2 cap are series resonant. But maybe it doesn't actually oscillate because of this.

I spotted the lack of any ESR in the long traces between amplifiers and this would have almost certainly caused negative resistance at the input. But it looks like you have since added parts to tame this. Not sure I want to look at it much beyond this other than to say you might find it has a gain bump up at UHF. Maybe the other stages mask this.

Your 20MHz amplifier looks to be very complicated but I assume you added all the complexity to achieve low noise down to low frequencies? Presumably you started out with the classic two transistor version of this feedback amplifier?  I've built the two transistor version a few times but only operated it down to just under 1MHz. But it typically runs up into VHF with the right transistors. I've built the two transistor version with BFR91 parts and achieved unconditional stability up into the UHF region according to a simulation and also a VNA. But I've only ever used the circuit as a bench/test amplifier. I'd expect it to be noisy if I built it with 2N3904 parts and it would struggle to work over a decent bandwidth.

I still have my BFR91 version if you want to see it? I can send you s-parameter data of the complete module up to 8.5GHz but obviously the amp only works up to VHF because I built it with leaded parts rather than SMD.
I can measure noise figure fairly accurately here with a reasonably good noise source and  the Y factor method. I can measure it for noise figure if that helps? Adding a choke in the right place helps a lot with noise figure but it spoils the LF response obviously.

[T3sl4co1l]

The 20MHz amp is only slightly more complicated than a two-transistor circuit, and at that, mainly because it's also not easy making a clean negative supply out of nothing...

Concept:
1. Take a bigass transistor.  Large junction gives low noise, small Rbb' and can operate at higher currents for smaller r_e.  (AoE3 measured ZTX751s very favorably, and go through a design process with them.)
2. Improvement: Ccb is huge, so cascode it away.  Lots more BW.
3. Increase gain (and a little BW) by replacing pull-up resistor with CCS.  Bias gets a little looser, but otherwise good.
4. I can use a base voltage divider (i.e., feedback resistor from emitter follower to base, then B-E resistor), but this is a bit sloppy for bias stability.  So I chose to get fancy and made a servo error amp.  Which pulls up, so, current mirror makes it pull down.  Four whole transistors just for a few tenths of a volt stability.  Well, I wouldn't say it's elegant, no.
5. I wanted to avoid three LF poles in the signal path.  There's one electrolytic at the input, and one at the output.  The servo amp is another pole too, but it needs much less capacitance, and it's out of the main signal path.

I initially had it with a fixed current mirror pulling down (to balance the feedback resistor's excess bias current), but that was very noisy (it picks up full supply noise, after all).

Incidentally, I was able to measure the same noise factor with the circuit built on solderless breadboard (using a breakout board and short leads for the PBSS303NX, or it might've been a PBSS4540 or something actually).  People constantly rag on them, but they're just as good as anything else in the right hands, and just as bad as anything else in the wrong hands.

Tim