Originally posted by WM6
View Post
Announcement
Collapse
No announcement yet.
RELIC HAWK
Collapse
X
-
Originally posted by WM6 View PostWorking under Ubuntu and I can't import your LTSpice file in my Qucs simulator. Depend on your demo for now.
1. LC tank circuit starts to return energy in source V3 because the cascode Q2&3 saturates.
2. Distortion of oscillation becomes remarkable.
3. Efficiency drops because tank pumping pulse has width more than 180 deg. For maximal efficiency we need conducting angle about 120deg (duty cycle about 1/3).
4. Control voltage Vo (the red line) should be very low (TL074 can not deliver such low level).Attached Files
Comment
-
Yes, I got the same, but I am not sure if in real world so little difference in power supply can lead to so huge distortion (a little unbelievable for 0.1V only)?
As you can see on second pic, the same happen with I circuit total I(V1).
Comment
-
Mike, you should remove the components in the red box. Unless I am mistaken, the 68k resistor was put there to isolate an oscilloscope lead during setup. One end is a test point.
The net segment that you labeled "Test" is connected to ground which means that 0V=test in your simulation.
Right click the net, and then "Highlight Net" and you will see what I mean.Attached Files
Comment
-
Originally posted by WM6 View PostYes, I got the same, but I am not sure if in real world so little difference in power supply can lead to so huge distortion (a little unbelievable for 0.1V only)?
As you can see on second pic, the same happen with I circuit total I(V1).
In the original circuit (see formulas in equivalent circuit diagram - post #957), to avoid distortion, we should adjust amplitude Vp no more than Vp = Vbat_min - Vsat = 10.2 - 3.8 = 5.4Vp, ie TX should generate voltage swing less than 10.8Vpp . This amplitude can be adjusted with resistors of voltage divider R6 ...R7.
However when battery voltage is 13.6V, the circuit oprates with very bad efficiency because theoretical possible swing is 27.2Vpp. TX power is proportional to square of amplitude Vp. The large saturation voltage 3.8V of cascode Q2&Q3 is gilti for this drawback. No need of cascode.
My idea is to remove Q3 and to use a reference voltage proportional to battery voltage. In the circuit below, Q3 generates very filtered such voltage. We can adjust R5 so that amplitude will be always almost equal to battery voltage despite its change.Attached Files
Comment
-
Originally posted by porkluvr View PostMike, you should remove the components in the red box. Unless I am mistaken, the 68k resistor was put there to isolate an oscilloscope lead during setup. One end is a test point.
The net segment that you labeled "Test" is connected to ground which means that 0V=test in your simulation.
Right click the net, and then "Highlight Net" and you will see what I mean.
Please click on emitter of Q1 to see voltage signal. The result is shown below. There is nothing interesting for measurement with oscope. Only a competent radio amateur will say that this is an inefficient TX circuit because the conducting angle of an amplifier class C is more than 180 deg.
As a redesigner of this TX circuit, I would connect an oscope to output of U1 because for me, the P.I. controller is the most important thing in this circuit, so it should be tested. The P.I. controller should operate:
a) quick (minimal settling time),
b) with wide change of output voltage Vo (without saturation of U1) and
c) stabile (without oscillation or ringing).
When I connect R10 to ground for a test, amplitude of oscillation increases. The controller senses this and starts to increase output potential Vo of U1 in order to decrease amplitude of oscillation to previous level. The oscope will show this settling process.
When I connect R10 to a positive voltage, amplitude of oscillation decreases. The controller should decrease output potential Vo of U1 in order to increase amplitude of oscillation and we will see on oscope if output saturates. To have very low voltage Vo without saturation, I use LM358 or LM324 instead TL074.
However in practice, to test settling process, I will connect R10 to a pulse generator with very low frequency, for example 2Hz and will see with oscope if the settling process is enough quick and if the controller operates with ringing. SPICE also can simulate this.Attached Files
Comment
-
Hi Olivier.
I made TX circuit of post # 937 for SPICE analysis by you with CircuitMaker 2000. It is attached here in compressed format. I have no time to make it for simulation with LTspice.
MikeAttached Files
Comment
-
Originally posted by o.mag View Posthi mike
a remark
in this diagram the ground is connected to the - terminal of the battery.
Ground use for the RX preamp.
do you want creates a negative voltage supplied to the other OPA ?
olivier
I think the TX has no enough power to supply increased looad because the negative rectified voltage is formed in half-wave when Q2 is not conducting.
As I mentioned in post #937, my idea is to supply only U2 (the RX preamp placed inside search head).
Comment
-
hi mike
OK : like this ! (the head is the red box )
olivier
Attached Files
Comment
Comment