AGC-controlled Termination Insensitive Amplifier experiment
ฝัง
- เผยแพร่เมื่อ 3 ส.ค. 2021
- The Termination Insensitive Amplifier (TIA) is a popular wide-band RF and IF amplifier, and a foundation of the Bitx, uBitx and many other superhet radios. A minor draw-back is its fixed gain.
In this video, I experiment with a mechanism (proposed by George VK4AMG) to add AGC to a TIA IF amplifier stage. The mechanism varies emitter degeneration and is similar to a scheme using PIN diodes shown in EMRFD Fig 6.41 (which is not a TIA).
This mechanism works, but there are side effects. Is it a useful hack? Tell me what you think!
Thanks to George VK4AMG, Bill N2CQR and Peter N6QW for helpful comments and feedback.
'A Termination Insensitive Amplifier for Bidirectional Transceivers', Wes Hayward, w7zoi, and Bob Kopski, k3nhi. © 26 June 09:
w7zoi.net/bidirectional_matche...
Facebook (VK Homebrew), scroll down to George VK4AMG's post, March 3rd 2021:
/ 1689037384702683
Blog post:
vk3hn.wordpress.com/category/...
Disclaimer: I am not an RF design engineer and I offer this as an interesting experiment to home builders and experimenters. If you want to challenge or debate the approach, I encourage you to model, build, measure and report your findings! - วิทยาศาสตร์และเทคโนโลยี
If you preceded that with a modest gain low noise amplifier (e.g., grounded gate JFET or something that won't compromise dynamic range), you could then follow that with a pad to stabilize the impedance mismatch. E.g., [LNA]->[PAD]->[TIA]. I've been playing with circuits based on EMRFD 6.32 and following for a direct conversion receiver, and your comments at the end reminded me of it. Probably crazy, but it was my first thought :-).
Yakov, that is in fact exactly what I am doing in the receiver I showed in the video (SP-11, under construction). It uses a BPF in front of a graounded gate J310 RF amp stage, then another BPF, into the DBM, a 2N5109 post-mixer amp, a 50 ohm pad, then into the TIA strip. And that's before I insert a diplexor, which I've been prototyping externally. After leaving it a couple of days and looking at the comments I am thinking that the AGC-controlled TIA is a useful building block, but you need to look at the impact of varying the load of the upstream receiver stage. As some have pointed out, that might not be significant.
EMRFD 6.32 looks very similar to SP-11. I used double tuned BPFs ahead of and after the J310. My post mixer amp is the classic 2N5109. You could use MPSH10s for the parallel pair shown, this is very similar to the approach taken by Eamon EI9GQ in RadCom/How to Build a Transceiver. I am about to insert a diplexor on the DBM's output port. Nice! Good luck with it.
Thanks for the video Paul. Nice to see someone trying out circuits. I have used similar circuits for gain control both at RF and audio. You can get good results by also using a FET in the same circuit. Hayward actually shows the use of 1 and 2 PIN diodes but they introduce distortion at low RF frequencies due to short charge time in the intrinsic layer (Ref Hayward, RF design page 255). The simplified gain equation for an un bypassed emitter is the collector resistor (or impedance) divided by the emitter resistor. So with a 220R collector resistor, and a 100R emitter resistor the voltage gain is 2. The transistor or FET or diode may put, say, a 10R equivalent resistor across the 100R which then gives a voltage gain of 22 times, which is around 27 dB. And going down to 5 ohms , gives a gain of 33 dB etc. However I found that as you increase the gain, the third order IMD gets worse and at a very low RF emitter impedance, the transistor starts to go from a linear class A RF amplifier to a less linear mode. Fiddling with the emitter impedance also lowers the base input Z. That is, it goes from a lot of negative emitter feedback with a high input Z (as in an emitter follower), to a lot less negative feedback, hence gain more gain, but lower input Z. So the amplifier input impedance goes up and down with AGC not a good situation if fed from tuned circuits or crystal filters. The best AGC controlled system is the AD603 IC which does not use variable degeneration but an internal voltage controlled resistor network resulting in low IMD at all values of AGC and with high dynamic range.
Lou, thanks for watching and your informative comment. Very interesting. One of the best things about posting this video has been the excellent comments it has solicited. I am in no way an RF designer, not formally educated in electronics, just a hobbyist, but I am sure learning quickly from the homebrew community. Not surprising that this mechanism has a number of instantiations and they are called out in the comments. Your comments on linearity, gain and input Z noted. I concluded that as an AGC in the TIA block, it is probably acceptable, as long at the preceding stage is not critically dependent on the varying impedance of the TIA block. In my receiver, the first TIA is driven from RF power from the first fioxe dgain IF amp via a 3dB 50r pad. The second TIA interfaces to the secondary (low impedance) winding on a crystal filter matching transformer, possibly a bigger concern. Thanks and please keep sharing your knowledge. -- VK3HN.
@@Paul_VK3HN Hi Paul, all good. Yes, try it out and see how it goes. IMD might not matter if it is not in a high level transmitter path since receive signals are not really that big till they exit the IF amps. Keep up the experimenting. Take care, cheers.
@@vk3aqz Tnx Lou. I've received an email from Wes himself, 😀 will make a followup vid soon. Speak soon.
Hi Paul. Thanks for another fascinating video. We've corresponded about this before as I used fixed gain TIAs in my recent Optimiser rig. As I actually daisy chain a couple of these modules on either side of my crystal filter I did wonder about putting some variable IF gain in-between each pair of TIAs - like a kind of AGC sandwich!
i.e. TIA -> Variable Gain IF Amp -> TIA -> Crystal Filter -> TIA -> Variable Gain IF Amp -> TIA
The TIAs would have to be built for pretty low gain so as not to overdrive. My thinking was that the TIAs would effectively buffer the connecting stages from any impedance changes whilst still allowing me some amount of AGC control before and after the filter. I haven't done any more than think about it yet but its on the TODO list! Thanks again. Great stuff! 73 Nick M0NTV
Hi Nick and good to hear from you again. I hope you might try this TIA hack sometime and let me know what you think.
I probably wouldn't mix and match them, rather go with one or t'other, the TIA is only one tool in the toolbox, even if it is gain controlled. Have a read of some of the comments, there is a suggestion for a very smart PIN diode voltage-controlled attenuator, which is yet another approach to IF gain control. And a couple of people chime in on places where this emitter degen idea is used for gain control elsewhere. Most interesting, I'm glad I got around to trying this. Have you moved yet? 73 from 3HN.
@@Paul_VK3HN Thanks Paul. We move next week so pretty much everything is in boxes right now! I'll keep following your videos though and get my homebrew fix that way for a little while. Keep up the great work my friend. 73 Nick
Very interesting. This approach is similar to EMRFD Fig 6.41, but instead of a NPN transistor authors choose to use a PIN diode as a voltage variable resistor (1N4148's work as well for low-level signals). The amplifier described there is not TIA though. I was thinking about building an IF derived AGC with TIA as well, and came up with another approach. Instead of changing the gain, it's possible to place a voltage variable attenuator on IF stage. I discovered that the circuit proposed by Raymond Waugh (Google for Waugh attenuator, it's easy to find) works well with 1N4148's, is well matched on HF (and up to ~150 Mhz if you don't mind SWR
Oh hey, I have been thinking about using something similar, but with PIN-diodes salvaged from TV tuners, as I still have over 10liters of them.
And using some 50ohm MMIC amps like MSA-0685 or MSA-0385 as the IF amps.
Excellent pick up Alex, thanks! EMRFD calls it 'PIN diode emitter degeneration', notes 30dB gain and 30dB swing at 10MHz, with PIN diodes, and that 'this circuit has promise'. I must have looked at this page a dozen times and never really noticed this. There is no C-B feedback resistor so the input inpedance relationship is different. I might try this in future.
I've tried primitive attenuation approaches such as Arduino-switching various pads across the frequency range. It hasn't always been effective.
As for Ray Waugh's invention at Agilent's App-Note AN1048, this is really interesting! So simple, and no more than a 1-hour prototype to build and evaluate. Great find Alex. Please comment here or email me if you try it out.
Thanks for your pointers, congrats on your FB multiband SSB/CW transceiver. Keep melting solder!
A PIN diode variable attenuator and MMIC gain stages would make for an interesting alternative to a traditional AGC-controlled MOSFET or bipolar IF strip, keen to hear if you try this!
The fact of the matter is that it worked, and we heard the result. So then what is the concern? Is impedance matching on receiver blocks really all that critical anyway? I don't know the answer, I've only scratch built one rig so far. But for the sake of argument, say you have an antenna preamp going into a diode mixer then this IF strip. The mixer in this case isn't changing the output impedance of the stage before it right? So a varying input impedance would look like a varying output load on the preamplifier, which would only mean the gain is changing. But you are also varying the gain of the IF amp to compensate... Seems it might be a wash. Does this make any sense? I'm kind of thinking out loud. :)
Hi Dennis, thanks for commenting. The layout you describe is the one I would be concerned about, that is:
[preamp] > [DBM] > [ TIA with gain control]. You don't want to vary the impedance of the load on the DBM's IF port, as DBMs are wonderful mixers but only when they are fed and terminated properly, so its strong signal and IMD performance may suffer. A better arrangement is to follow the DBM with a fixed gain post-mixer amplifier, often in class A for best linearity (classic RF transistors like 2N3866, 2N5109 are often used here), then a 3 or 6dB 50 ohm pad, then the gain-controlled TIA. In that arrangement, the TIA interfaces to a fixed pad, so the only consequence is a drop in power transfer due to impedance mismatch as the TIA gets gain-clamped. Which as you point out, is what you are doing in the rest of the IF strip anyway! 73 and keep melting solder.