FranzHere's what my LTSpice analysis of the "Ver 2.0" BiDi amplifier looks like (measured at 20 mhz):
Gain:~= 20 db Input match: ~= 7.5 db Output match: ~= 30 db Standing current: ~= 50 ma Noise figure: ~=2.8 dbYou can improve the input match by changing R11/R19 to 5.1 ohms and R10/R64 to 1000 ohms. This will result in a 50 ohm match of about 20 db with very littler gain change.
DC performance is another story. The amplifier idles at about 50ma, which is far more than necessary. The culprit here is the emitter follower output stage. Emitter followers are great when driving purely resistive loads, but they tend to fall down when trying to drive large voltage swings into capacitively coupled loads. This is because the follower can source current but it can't sink it. For this reason, designers put small resistor values from the emitter to ground to improve sinking. This also raises the bias current for a given bias voltage. The 220 ohm output resistor was undoubtedly selected for this reason. A better solution with almost the same number of parts would have been to use a complementary NPN/PNP output stage which would cut current consumption by at least 3x and improve the ability to swing AC voltages into the capacitively coupled load.Also, the collectors of Q1/Q4 are biased at 10.5V, which leaves very little headroom for strong signals given the 12V supply.
As to your 30 mhz xtal filter, I think that's a move in the right direction for this radio. The problem I had was that the crystal losses at that frequency were so high that a 'proper' bandwidth (~2.5 khz) would have far more loss I think would be acceptable. In any case, I find that ordering crystals from a reputable source (I use ECS crystals ordered from DigiKey), results in a much smaller spread of frequencies and generally better crystal Q and loss. For my 30 mhz filter, I just picked 8 crystals out of the bag, built the circuit and it worked pretty much as predicted. This is also my experience with all of the xtal filters I built in the last few months (about 10 of them). If you buy good crystals, the only thing you really have to match is the frequencies, and they'll be very close for SSB purposes if you buy from a known reputable source.
Joe W3JDR----- Original Message ----- From: "Franz DX" <dd4wh.swl@xxxxxxxxx>
To: <minima@xxxxxxxxxxxxx> Sent: Thursday, November 20, 2014 3:22 AMSubject: [minima] Re: Preliminary feedback on Minima version 2.0 Farhans mail Nov 1st 2014
Hi Joe, thanks for the filter matching hints, I will build a narrower 20MHz filter with matching LC-networks. I also now have 30MHz crystals, but I did not yet start building, because bandwidth for a 50Ohm filter impedance was calculated by Dishal as 7kHz, which is much too wide. As I now know how to match a filter with narrower bandwidth and lower impedance, I can proceed with that. IF amp: As I understood, the topology of the amp is that of Wes Hayward from 26 June 09. http://w7zoi.net/bidirectional_matched_amplifier.pdf In his paper, there is a nice table with different resistor values for load (Rl), feedback (Rf) and emitter degeneration (Rd) for establishing different gains and Input RL. So now I tried to summarize the different values for the different versions of Minima (given in the order: Rl, Rf, Rd) from the two Minima circuit versions and the list: original Minima: 220, 1K, 10: gain 17dB, InpRL: 16dB Joe Rocci (on the list on 15/10/2014): 1K, 1K, 20: gain 20dB, InpRL 22dB Minima v2.0: 220, 2K2, 10: gain ???, InpRL: 7dB One further version from Wes´ paper could be: 330, 1K8, 4R7: gain 24.3dB, Inp RL 23dB So, for me, as a beginner in homebrewing, the question remains: which one should I build ? What puzzles me, is that the versions also differ in the value of the emitter resistor of the second transistor of the darlington emitter follower: Farhan uses 220ohms, Wes uses 470ohms: does this make a difference in gain, output impedance or other features? And Joe: which value did you use for that resistor in your LTspice model? Thanks in advance for your comments! 73 Frank DD4WH 2014-11-19 18:43 GMT+01:00, Joe Rocci <joe@xxxxxxxxxx>:
Franz You could ignore the filter mismatch, in which case you'd incur anadditional db or so of passband ripple and some added mixer/filter loss, or you can match the filter properly. To do this, you can build an "L" matching network on each end of the filter to match 25 ohms to 50 ohms. This would be a series inductance of 0.2uh and a shunt cap of 160 pf on the 50 ohm side.Almost anything that gives the correct inductance will work here, as the loaded Q is very low and the circuit is forgiving. An example toroid inductor would be 7 turns on a T50-6. Also note that the BiDi IF amplifier in the Ver2.0 schematic is not really optimized. The 50 ohm input match is only about 7 db and the bias point is way off from optimum, resulting in twice as much standing current as necessary and very little voltage-swing headroom for strong signals. In areceiver that doesn't have IF AGC, it's very important that the IF have lotsof dynamic range, so this should be addressed. I addressed it a while ago, but my suggestions didn't get incorporated. Joe W3JDR ----- Original Message ----- From: Franz DX To: minima@xxxxxxxxxxxxx Sent: Wednesday, November 19, 2014 10:15 AMSubject: [minima] Preliminary feedback on Minima version 2.0 Farhans mailNov 1st 2014 Thanks Farhan, for your work on Minima optimization! Here are some of my experiences from building some of the modifications Farhan implemented in his mail from Nov 1st 2014. In order to be explicit with the version numbers, I will call that „Minima v2.0“, although Farhans schematic calls it Rev1.0.1.) Mixer with FST3253 works well. I built it first with the double balunon a binocular core by Joe Rocci, that worked well on receive. But then Irealized that Farhan meant the simple Guanella balun, which I built using an FT37-43 with ten bifilar windings. Good receiving results with that mixer, Idid no measurements, however.2.) I still use my old crystal filter with 20MHz crystals and 100pF caps,that has a 4.5kHz bandwidth. Farhan, you changed the capacitors in thefilter from 100pF to 200pF, probably because you want a narrower bandwidth.But with my crystals, that would end up with an impedance of 25-30 Ohmsaccording to calculations of the QER filter calculation module of the Dishalprogram. So does the mixer guanella balun cope with that impedance? And: Isnt there a mismatch between the IF Amp and the filter also? Or did you change the values of the IF amp in order to cope with that lower impedance of the filter with 200pF capacitors and thus narrower bandwidth? 3.) IF amp has some changed values. I am not experienced enough to use Wes´ calculation hints to calculate gain, reverse isolation, impedance IN/OUT and NF for that amp. Can anybody help me with that or has anybody measured the amp? 4.) I built the audio filter after the two-diode-mixer: It heavily picks up hum and additionally click sounds from the digital board. When stations are strong, this is not an issue, but with weak stations, you can hear clicks and hum. So, probably shielding the audio filter would help? I have not yet tried the other audio modifications. Would appreciate comments and/or helpful hints on my experiences. Best wishes from rainy Dresden, Germany, DD4WH Frank