One trick is to add a big resistor and see how it affects the frequency rolloff, and calculate capacitance from that.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
I don't know why I can't have just a few days to experiment with this amp. BTW, here is the manual. >
Page 3 has some specs, one is the input R and C. At 500kHz it's 122k? with a parallel 37pf capacitor. At 1700kHz it's 52k? with a parallel 37pf Capacitor.
Typical input intercepts are: IIP2 +87 to +93 dBm; IIP3 +40dBm to +45 dBm.
3db frequency range is 50 kHz to 300 MHz.Not shown is the noise figure.
I think the first time I applied power to it was 4 days ago. I haven't listened to it more than 60 seconds. The original question was not even about the amp, it was about designing a transformer to feed a high impedance load.
Seems I now have built a transformer and have a question about it.
I have a 10 turn to 40 turns transformer on a #43 material, binocular core. I realize the transformer is a rather high impedance for RF, but I'm here to learn what the complications are. If I put a 12,100? resistor on the 40 turn winding and scan the Primary at 1MHz, I get R=336? and X= -292?. The R is right on, the Reactance is capacitive and to high. Is there anything I can do to make this less reactive? A wider range sweep has a downward slope to the R but the Reactance is fairly constant. Sweep shown here.
Mikek
Oops. You're right. The phasing dots should be on the same end of the transformer.
Oops 2.0. You're right again.
The voltage is set by a resistor divider made of two 1M resistors. In order to produce 5.2v instead of the theoretical 6.0V, the voltmeter would need to have an input resistance of: 5.2/12 = 0.433 To get this ratio, the grounded 1M resistor should be: 0.433 * 2M = 0.866M To obtain 0.866M, a meter resistance in parallel with 1M would be: 0.866M = (R * 1M) / (R + 1M) Therefore R (meter resistance) = 6.5M
I couldn't find any ohms/volt specs for my Extech/RS 22-816 multimeter. So, I measured it. I put a 10V power supply and a pile of high value resistors in series with the DVM input. When the meter read 5.0V, the meter resistance is equal to the resistor value. I strung 10ea 1M 1/4w 5% carbon film resistors in series and ended up with 4.95V. Therefore the meter resistance is about 10M. It's not my predicted 6.5M but with all the probable errors involved, is close enough.
Since I had the 1M resistors handy, I put two in series to form a divider similar to the RF amp circuit. With the DVM across the grounded resistor and 12V applied, I got 5.48V, which is close enough to the measured 5.2V.
So, you're right. It was meter loading that caused the voltage error and not a blown J271.
My background is in marine radio design, where equipment is expected to operate normally when somewhat wet. Under some conditions, G10/FR4 is a can simulate a sponge. Therefore, low impedance and resistance design is the norm. Every time we used resistors higher than about
10K, we got into trouble with board leakage and water related problems. I think 100K was deemed the largest value considered acceptable, although we did use larger values when desperate. I would never have used 1M resistors in anything. With low resistances, meter loading isn't even a consideration (although capacitive loading on RF circuits is a consideration). My apologies, but after years of living in a low impedance world, I just didn't think about the possibility that the meter would affect the measurement.In this design, the antenna impedance is about 400 ohms. 10 times that is considered sufficient to prevent the JFET bias resistors from having any influence on the antenna loading. Therefore, the resistors could have been as low as 80K and not had an effect. Optimum NF (noise figure) is not an issue in the BCB (broadcast band) where the atmospheric noise is so much higher than front end receiver noise[1].
[1]Except if one is designing a MiniWhip RF amp, where the antenna impedance is so high that 1M to 10M bias resistors are common:-- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
You might be right. I didn't notice that.
The source voltages for both JFET's on the original schematic show 7.62V and 5.35V, which is roughly what I would expect for maximum output swing from the source followers. The average is: (7.62 + 5.35) / 2 = 6.49V which is close enough to 6.0V to work properly. It would be tempting to split the 220 ohm resistor in two with two 100 or 120 ohm resistors, and adjust the potentiometer for 6.0V at the junction of the two resistors. However, that assumes that the two JFET's are reasonably well matched. If that's not the case, then a 2nd potentiometer might be a useful addition.
-- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
So this is some kind of Beverage with a 341 ohm termination resistor at the far end, which explains the comparatively flat resistance and reactance curves. If the interest is mainly reception in the MF broadcast band, you could even tune out the capacitive reactance.
But the real question is, why do you need the source follower amplifier at all ?. Why not just put a 9:1 impedance transmission line transformer between the antenna and coaxial cable (and 50 ohm receiver input) ? Alternatively use 4:1 transmission line and 100 ohm balanced feedline. No need for some amplifiers,
On LF/MF the band noise is so strong that with a half decent antenna, it will mask the receiver input noise.
Unfortunately Beverages have a gain much worse than 0 dBd. Only ferrite rods are worse with gains of -40 to -60 dBd, but still usable thanks to the high band noise.
One alternative would be to put just a matching transformer between antenna and feedline and if the receiver has a bad sensitivity at LF/MF, use a manually tuned indoor preselector followed by 10-20 dB gain. The preselector will reduce the risk of overloading the receiver.
The source follower amplifier with 330 kohm input impedance is intended for some random wire antenna, which is much shorter than 1/4 wavelength and hence gas a highly capacitive reactance.
Yes, I have that, with a little complication, I match the antenna 330? to a 100? feedline, then at the end of the feed line, I have a 100? to
50? matching transformer to the radio.You can follow the signal path through the relays and transformers here.
Note: if I don't power the circuit the amp is not used. This makes for a very simple A/B test of amp vs transformer.
and if the receiver has a bad sensitivity at
On page 15 and 16 of the manual it list some antennas that this was designed to work with.
If you need articles on those antenna, I may be able to find them.
Mikek
It's pretty reasonable for preamps in this class. Remember that the IP3 is the theoretical IMD intercept, not a valid operating point.
Lankford and Jack Smith at Clifton Labs both had a lot of experience with this sort of thing.
-- john, KE5FX
I'm quite aware of that. However, IP3 is commonly within 20 dB of the
1-dB compression point even for a very linear amplifier, which for 10 mA of bias current is not +28 dBm, by a lot.The claimed IP2 is the really ludicrous one. It requires a level of matching between the N- and P-FETs that beggars belief.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
I don't know if you saw my post of the manual, It has some specs on page 3 and says,
typical input intercepts are: IIP2 +87 to +93 dBm; IIP3 +40dBm to +45 dBm
You can believe it or not, I believe the numbers will be high, that's is one of the main criteria for a BCB preamp. There are so many signals you do everything to avoid mixing them.
2nd and 3rd order intermodulation measurements on page 11/12.Mikek
So they claim. With input signals at 0 dBm, an amp with +93 dBm IP2 would generate second order products at -93 dBm as well. That requires matching of the discrete FETs' transconductance vs I_D to a few parts in
10**5 at signal levels near the amp's P_1dB.Maybe in the SPICE spherical-cow universe.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Around 0 dBm is way too much input power for measuring IP2/IP3.
3rd order intermodulation products rise by 30 dB for an input power increase of 10 dB. This is simply impossible for an amplifier approaching saturation.Therefore one gets results that are _much_ too good.
IP must be measured at levels where the intermodulation products just come out of the noise.
cheers, Gerhard
0 dBm into 50 ohms is about 220 mVrms and 300 mVpk. The worst sum voltage is 600 mVpk or 1.2 Vpp, which is the required linear voltage swing.
At +12 Vdc, the available swing might be 9 Vpp or 3 Vrms or about +20 dBm, so the P_1dB is slightly larger than this.
After all, this is a voltage follower. At least for bipolar common collector circuits the voltage gain is often quoted as 0.99.
The distortion is created due to the curvature of the transfer function. How much curvature does a voltage follower have well inside the supply voltage limits ?
Does a JFET common drain vollower behave differently from a bipolar common collector voltage follower ?
I'm still looking for what I can change to get a less reactive transformer. I have ordered a higher permeability core material.
Mikek
You aren't getting 3V RMS with 10 mA of bias current. Remember, dB is a power ratio, not voltage.
-93 dBc is 0.002% THD. You ain't getting that with an unbalanced complementary follower in large signal conditions with no feedback.
Depends on the load, but with a JFET with no drain bootstrap, it has lots. The drain impedance of JFETs is the pits, and yours are mismatched. Second-order distortion comes from the two half-cycles being amplified differently. At any given frequency and one choice of signal levels, there might be a pot setting that nulls out the second harmonic, but that's a cheat--anyplace much away from that, the performance will go back into the tank.
Intercept point is not measured at a single input level--you start from a low level, and plot the spur amplitude vs. input level. At low levels, an Nth order product will go as the Nth power of the input levels. You draw the straight lines on a log-log plot, and the intercept point is where the extrapolated lines intercept. Saying that the intercept point is measured 'at 0 dBm' is complete bollocks, and strongly suggests chicanery, as explained above.
Yes, it's worse because of the low drain impedance.
With a bootstrap and some local feedback to keep the JFETs' bias conditions constant, you can do much better.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.