What is a noise-blanker?
A noise-blanker, the famous NB, should act as a fast gate and when a noise-peak is detected this NB-gate is locked. During the noise-peak there is no signal to the demodulator. Since most noise-peaks are very short, you will not hear that the receiver is "quiet" during this gate-time and you will still be able to hear the normally received background noise. The only difference is that the noise-peak is filtered if the NB is working properly.
How the NB is electrically made in the FT225
In the FT225 this NB-gate is a diode in the "RX RF unit", D09, originally a 1S1007. When there is a DC-current through this diode, there will be a signal going through to the "SSB IF unit". (There is not any NB-function when FM-mode is selected.)
In the "RX RF unit", or the Mutek-board, the signal from the mixer is splitted to two amplifiers made of Q04 and Q05, (2SC784R).
One of these amplifiers feeds the NB-gate, and later this signal is fed to the "SSB IF unit". The other amplifier feed the "FM IF unit".
In the "FM IF unit" there is a noise-amplifier and after this a peak-detector made of two diodes. The noise-peaks are amplified and fed to the gate-driver, Q04 in the "FM IF unit".
When the NB-gate is activated the output from the gate-driver is low, nearly 0V. In the original "RX RF unit" the NB-gate close at about 1,8V, the Mutek at about 4,0V.
One reason why the noise-blanker is so bad is that the voltage on the base of Q04, "FM IF unit", must be about 0,7V before the gate-driver closes the NB-gate. The base of Q04 is originally biased at about 0,25V and when a pulse is present this voltage raise and sometimes the gate-driver close the NB-gate, ie if the voltage on the base of Q04 is above 0,7V.
By increasing this "standby-voltage" to about 0,4V more pulses will close the gate! I made this modification by changing the R10, "FM IF unit", from 100K to 56K.
In the "RX RF unit" I also changed the C38 from 47nF to 1nf. I think that this speeds up the NB-gate and it will cut of the signal faster. Later I also increased the voltage on R31, "RX RF unit", from 8V to 13,6V. (This is also made in the Mutek-board.) In this way the control-voltage from the gate-driver only has to drop to about 4,0V before the NB-gate close. The gate-driver output is normally about 6,2V when there is not a noisy input.
By doing this modification the NB works much better. I have still not tested the NB in a large contest with "mega-signals" on the band. There is a possibility that the NB could cut these signals with heavy distortion as a result.
One can play around with different values on the R10, "FM IF unit", I only measured the output from the gate-driver and it looks as if it now stops most noise-peaks.
Why the NB can not be better
This circuit is far from perfect. One reason is that the signal must pass several amplifier-stages before "decision" is taken to close the NB-gate. The basic error is that the signal from the crystal-filter basically feed the NB-gate at the same time as it feed the noise-detector! If there was a delay before the NB-gate, then the detector would have some time to make a decision and more peaks could be attenuated properly! With the present configuration most noise-peaks will go trough the NB-gate, maybe not the hole peak but at least the beginning of the peak until the noise-detector make a decision to close the NB-gate.
The idea to improve this system further is to take the signal to the noise-detector before the crystal-filter in the front-end. In this way the noise-signal will get faster to the noise-detector, or at least before the signal has arrived to the noise-gate! The signal to the noise-detector will also be more broadband.
If more delay is needed an extra LC-filter as a bandpass-filter in the SSB/CW path, ie before the NB-gate, could delay the signal further. I have seen this in other receiver designs, so this is nothing new! I will get back to this in my new front-end design.